<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[lattice-automation - Medium]]></title>
        <description><![CDATA[We make software solutions that enable synthetic biology workflows, pipelines, and best practices. We use experimental data and institutional knowledge to enable bio-design productivity and learning. - Medium]]></description>
        <link>https://medium.com/lattice-automation?source=rss----736e74318f86---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>lattice-automation - Medium</title>
            <link>https://medium.com/lattice-automation?source=rss----736e74318f86---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 01:17:59 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/lattice-automation" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Improving Codon Optimization with Recurrent neural networks]]></title>
            <link>https://medium.com/lattice-automation/improving-codon-optimization-with-recurrent-neural-networks-49655ee8a7f5?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/49655ee8a7f5</guid>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[synthetic-biology]]></category>
            <category><![CDATA[biotechnology]]></category>
            <category><![CDATA[genetics]]></category>
            <category><![CDATA[recurrent-neural-network]]></category>
            <dc:creator><![CDATA[Lattice Automation]]></dc:creator>
            <pubDate>Thu, 20 Apr 2023 02:00:23 GMT</pubDate>
            <atom:updated>2023-04-20T02:00:23.586Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/535/1*uvCAnoBCa0zJZiYE28_tQQ.png" /><figcaption><a href="https://www.tools.latticeautomation.com/">tools.latticeautomation.com</a></figcaption></figure><h3>Background</h3><p>Recombinant DNA technology has provided cardinal breakthroughs in biomedical research — from cheap insulin production to hepatitis vaccines. As of 2022, the recombinant therapeutic market was valued at over 2.8<strong> billion dollars</strong> and expected to double by <strong>2029</strong> (Source: <a href="https://www.globenewswire.com/en/news-release/2023/02/08/2604081/0/en/Recombinant-Protein-Market-to-Surpass-US-6-691-4-Million-by-2030-Says-Coherent-Market-Insights-CMI.html">Coherent Market Insights</a>).</p><p>The central dogma of molecular biology states that <strong>DNA</strong> is transcribed into <strong>RNA</strong>, which is then translated into <strong>proteins</strong>. This idea has been the foundation of research in recombinant technology, allowing us to design plasmids (see another Lattice tool,<a href="https://medium.com/lattice-automation/repository-based-plasmid-design-312744dd484d"> repp</a>, for DNA assembly) and express proteins of our choosing.</p><p>A variety of factors influence expression levels of recombinant proteins, including the choice of expression vector, promoter, and <strong>codon bias</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TPjFk-MYp8m1ySLTltFTsQ.png" /></figure><p>Optimization without considering the codon usage or codon bias of the chassis is not ideal! Let’s take a look at codon optimization… (“Running Away Balloon Meme” format from<a href="https://knowyourmeme.com/memes/running-away-balloon"> Superlmer</a>)</p><h3>Codon Optimization</h3><p>Every organism has its own codon bias — or the preference at which it uses codons in its genome. Since translation of mRNA into proteins is a process that ultimately relies on the codons present in the mRNA, understanding codon bias can help us improve expression levels of recombinant proteins.</p><p>This is the idea behind codon optimization, which involves changing the codon sequence of the target gene to match the optimal codon usage of the expression host. In this way, codon optimization can significantly boost expression levels of recombinant proteins. In an <em>Escherichia coli</em> (<em>E. coli</em>) chassis, increases of two to fifteen-fold have been observed!</p><p>This is pretty exciting, and it is core to the ideation process for our recent release of ICOR — a codon optimization tool that uses AI to best optimize genes for expression in <em>E. coli</em>.</p><p><em>Check out ICOR’s accompanying GitHub and Publication</em><a href="https://tools.latticeautomation.com/"><em> here</em></a><em>!</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SO35Bv2H0VCGEJPSFSjIPw.png" /><figcaption>User workflow for sequence codon optimization using ICOR deep learning model with overview of model creation. (A) A user workflow towards creating a vector for heterologous expression is depicted. (B) Expanding out of “sequence codon optimization” in the user workflow, the overview of the ICOR model creation is given. In a production setting, a trained and packaged model of ICOR is inferenced.</figcaption></figure><h3>Why AI?</h3><p>It’s evident that artificial intelligence (AI) technology has been shaking up everything around us. It’s no surprise that AI may offer new insights into the codon optimization problem.</p><p>Our review concluded that many industry-standard codon optimization tools rely on biological indexes that replace synonymous codons with the most abundant codon found in the host organism’s genome.</p><p>The codon CGU (corresponds to Arginine) is used about 42% of the time for Arginine in <em>E. coli</em>. Many codon optimization tools would replace <em>all </em>codons in a theoretical sequence with CGU. This is statistically unlikely to be the best choice given the hundreds of thousands of permutations available.</p><p>More advanced, are frequency-based tools. These will ensure that, after optimization, 42% of the codons for Arginine in the sequence would be CGU. To some extent, this can relieve the metabolic stress of expression that would be caused by a ‘one amino acid–one codon’ approach. However, even frequency-based tools don’t fully address the energetics of translation.</p><p>The context of specific codons is also essential to the translation process. For example, the codon CGU may find itself being close neighbors to certain codons. Or, patterns of codons may appear — context that can be difficult for traditional algorithms to identify on a genome-wide scale.</p><p><em>These technical (and biological!) aspects are discussed further in our publication.</em></p><p>Artificial intelligence, more specifically, recurrent neural networks (RNNs), can be used to detect codon context across a large sequence.</p><h3>The Recurrent Neural Network</h3><p>Recurrent neural networks (RNNs) are designed to detect temporal patterns. As such, they are perfect for analyzing codon context. This was the method behind our ICOR optimization tool, which we trained on a dataset of thousands of genes from <em>E. coli</em>.</p><p>We utilized the Bidirectional Long Short-Term Memory (LSTM) architecture to train the model. BiLSTMs are a form of RNNs, which, as the name suggests, may help preserve context both forward and backward in a sequence.</p><p>In training such a model, we slap a layer of natural language processing. Similar to how words construct sentences, codons construct a DNA/RNA sequence. Therefore, the input to the LSTM model was represented as a sequence of codon embeddings — performed via one-hot encoding methods.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z4txrrSuYWpbfFcXQ8ZWuQ.png" /><figcaption>A full, deep dive into the ICOR neural network. More details in our accompanying publication.</figcaption></figure><h3>ICOR</h3><p>ICOR is an open-source software we have created that uses deep learning (see above) to learn codon usage in <em>E. coli</em>. We compile a dataset of over 7,000 non-redundant, high-expression, robust genes which are used for deep learning. Then, the model is trained to optimize sequences for high expression and to match codon usage in <em>E. coli</em>.</p><p>We test ICOR against several other codon optimization approaches, and deem that its performance is superior. The expression levels of our optimized sequences have an estimated 236% increase in mRNA expression compared to unoptimized genes.</p><p>Our tool is provided as an open-source Python package and is available on our accompanying<a href="https://tools.latticeautomation.com/"> GitHub repository</a>.</p><p>We also establish a benchmark dataset for codon optimization, providing researchers a reliable way to evaluate the efficacy of their models.</p><h4>Usage</h4><p>ICOR (requires Python) can be installed in just three lines of shell script:</p><p><strong># Install package</strong></p><pre>git clone https://github.com/Lattice-Automation/icor-codon-optimization.git</pre><p><strong># Install prereqs</strong></p><pre>pip install -r requirements.txt</pre><p><strong># Run ICOR optimizer</strong></p><pre>python ./tool/optimizers/icor_optimizer.py</pre><p>Our command-line codon optimizer accepts FASTA sequences as an amino acid or codon sequence. It will then return an optimized sequence for expression in <em>E. coli</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hpWwjxEAMKKY6K6h2V2uYg.png" /><figcaption>ICOR command-line script for optimizing sequences!</figcaption></figure><h4>Additional Tools &amp; Scripts</h4><p>Our software package also contains useful scripts for evaluating and testing out codon optimization techniques. We provide the following:</p><ul><li>5 different optimization approaches to test (Background frequency choice, ICOR, Extended random choice, Uniform random choice, Highest frequency choice)</li><li>CDS conversion — a script that takes an input of DNA sequences and fetches their CoDing Sequences from the NCBI’s nuccore database.</li><li>Benchmarking — an interactive Jupyter notebook for benchmarking the optimization methods on FASTA sequences.</li><li>Codon map — a codon map for amino acid to codon conversion in a Pythonic framing (useful for future projects in this space!)</li><li>E. coli’s codon frequencies — the frequencies and weights of each codon/amino acid in <em>E. coli.</em></li></ul><h3>Final Thoughts</h3><p>By understanding and applying the context of codons, AI-powered codon optimization programs, such as ICOR, can push the boundaries of recombinant expression.</p><p>We imagine this tool being used in research and industry! Researchers can take advantage of an improved yield towards designing more viable proteins, and the industry can utilize ICOR for cheaper, efficacious yields.</p><p>We’d love to see other scientists and engineers collaborate with us on improving the ICOR approach. That’s why it’s fully open-source. Feel free to open up an issue on our<a href="https://github.com/Lattice-Automation/icor-codon-optimization"> GitHub repo</a>, submit a PR, or even leave a star :)</p><p><em>Written by: Rishab Jain</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kKpZAyls4HGoMnIpR4jbQg.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49655ee8a7f5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/improving-codon-optimization-with-recurrent-neural-networks-49655ee8a7f5">Improving Codon Optimization with Recurrent neural networks</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Lattice Automation: Ferraris Not Fords]]></title>
            <link>https://medium.com/lattice-automation/lattice-automation-ferraris-not-fords-31e1761809fc?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/31e1761809fc</guid>
            <category><![CDATA[biotechnology]]></category>
            <category><![CDATA[software-design]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[synthetic-biology]]></category>
            <category><![CDATA[software]]></category>
            <dc:creator><![CDATA[Lattice Automation]]></dc:creator>
            <pubDate>Wed, 07 Apr 2021 16:27:26 GMT</pubDate>
            <atom:updated>2021-04-07T16:27:26.702Z</atom:updated>
            <content:encoded><![CDATA[<p>When you think about cars, I am sure you have some image in your mind of a Ford. You can also imagine a Ferrari. In addition to the images, there are probably some basic responses you have when asked about their affordability, quality, diversity, history, performance, and public image. You certainly know both companies. You know they make cars. You know they are both important. You also know they are not the same.</p><p>Started in 1903 by Henry Ford in Detroit Michigan, the Ford Motor Company was one of the first automobile companies and is known for its innovation involving assembly lines to produce automobiles¹. The assembly line was crucial in mass-producing automobiles and making them affordable and available for the middle class. The most famous example is the “Model T”, selling over 15 million units after being first produced in 1908. Ford makes something broadly used, reasonably priced, and targeted for average performance.</p><p>Ferrari was founded in 1939 in Italy, but not recognized as a true automaker until 1947. Ferrari was explicitly created to make race cars and was not targeting cars for mass consumption². Over the years, Ferrari has created some of the most expensive and high-performance cars on the planet³. The 1963 250 GTO sold for over $70 million and the SF90 is able to accelerate from 0–60 mph in less than 2.5 seconds. As of the end of 2019, the total number of Ferraris built and sold in their entire history is only 219,062. Ferrari targets a narrow market segment, prices based on expertise, materials, and brand, and is engineered for the highest performance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/599/1*2pJaRgDBF75Q5gm-1ejtIQ.png" /><figcaption><strong>Figure 1: Automobile Innovation.</strong> Early Ford assembly lines made cars widely available and affordable. Ferrari designs cars specifically for high-performance customer needs and focuses explicitly on creating cars with a level of performance unmatched in the industry. [LaFerrari sketch by design director Flavio Manzoni.]</figcaption></figure><p>Both of these companies make cars. However, they have fundamentally different approaches to design, manufacturing, and promotion. Both are highly successful and important contributors to automobile history. It is clear, however, that no one confuses one for the other and customers rarely are deciding between the two. <strong>One is not strictly better than the other.</strong> For example, I would not buy a Ferrari to commute to work, tow a boat, or get the kids to a soccer game. On the other hand, I would not use a consumer-level Ford to demonstrate racing prowess or to use it as an example of high-end design techniques. <strong>The overall message is that Ferraris are intended for specific use cases and users, not intended for general consumption, and cultivate a very special relationship with their customer base.</strong></p><p>It is in that way that Lattice Automation attempts to position itself in the world of bio-design automation software. In the design of biological systems, there are many tasks that need to be completed in the “specify, design, build, test cycle”. Some of these operations are highly applicable to large groups of people. These include viewing DNA sequences, designing primers, searching for sequences, laboratory information management systems (LIMS), recording experimental data, electronic notebooks, and storing protocols (just to name a few).<strong> </strong>Software is needed for these areas and is highly valuable.<strong> However, making universal software solutions that will apply to a generic customer base is not what Lattice does any more than Ferrari designs cars to be used by commuters.</strong></p><p>Instead, Lattice works with specific customers to create high-performance, highly specialized software. Not only is the code customized, but the algorithms, license terms, and ownership are all tailored to the specific needs of the customer. Lattice does not expect software to be available “off-the-shelf” to fit these customer needs. Furthermore, Lattice does not expect customers to gain any competitive advantage by using available off-the-shelf software. Lattice is unique in that it revamps biological design by tackling niche biological problems using novel software solutions and capitalizes on opportunities, informed by modern research and technology, that others may not see. Lattice helps customers create their “Ferraris” to complement the “Fords” that their organization already has access to.</p><p><strong>This doesn’t mean that Lattice will not create software that can be used by a larger customer base.</strong> Rather, it is not the focus. Ferrari has created design techniques that have been incorporated into the larger automobile industry. Lattice will do the same by providing open-source software “building blocks”⁴ that can be used as part of larger custom software engagements⁵. These tools can start future customized engagements, empower groups without access to closed commercial software, or allow us to partner with other software providers on larger projects. In the future, there may be Lattice technology in all major bio-design software offerings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ddd1zGd_raExXKG1XDLC4Q.png" /><figcaption><strong>Figure 2: Investment in research and development. </strong>Ferrari spends 18.8% on R&amp;D as a percent of sales (more than any other car manufacturer)⁶. Similarly, Lattice works on research and development projects internally to continue to innovate in the space, create intellectual property, and build its brand. [tools.latticeautomation.com]</figcaption></figure><p><strong>A customized customer engagement model also is a strategic decision</strong> whereby Lattice explicitly engages clients to cover the biotechnology industry in very targeted ways. We create symbiotic partnerships and enable businesses that will further the computational focus of bio-design to create future opportunities for the entire industry.</p><p>The 21st century promises to be driven by biotechnology. The design of biological systems using computer software will clearly be a major part of that revolution. Lattice has positioned itself to be the leader in high-performance, customized, highly innovative bio-design software. This comes from our close collaboration with some of the industry’s most impactful participants. We believe this engagement style will push bio-design automation forward while simultaneously leaving a generalized framework for tools to follow. We are excited about this process and hope you chose to reach out to us to collaborate on this journey together.</p><p><strong>References:</strong></p><ol><li><a href="https://en.wikipedia.org/wiki/Ford_Motor_Company">https://en.wikipedia.org/wiki/Ford_Motor_Company</a></li><li><a href="https://en.wikipedia.org/wiki/Ferrari">https://en.wikipedia.org/wiki/Ferrari</a></li><li><a href="https://digital.hbs.edu/platform-rctom/submission/ferrari-innovation-on-the-road-and-track/">https://digital.hbs.edu/platform-rctom/submission/ferrari-innovation-on-the-road-and-track/</a></li><li><a href="https://tools.latticeautomation.com/">https://tools.latticeautomation.com/</a></li><li><a href="https://www.ginkgobioworks.com/2020/02/19/accelerating-dna-design-software-development-with-seqviz/">https://www.ginkgobioworks.com/2020/02/19/accelerating-dna-design-software-development-with-seqviz/</a></li><li><a href="https://www.knowledgeleaderscapital.com/2019/05/14/spotlight-ferrari-under-the-hood-at-the-worlds-most-innovative-car-company/">https://www.knowledgeleaderscapital.com/2019/05/14/spotlight-ferrari-under-the-hood-at-the-worlds-most-innovative-car-company/</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=31e1761809fc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/lattice-automation-ferraris-not-fords-31e1761809fc">Lattice Automation: Ferraris Not Fords</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Repository-based plasmid design]]></title>
            <link>https://medium.com/lattice-automation/repository-based-plasmid-design-312744dd484d?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/312744dd484d</guid>
            <category><![CDATA[gibson-assembly]]></category>
            <category><![CDATA[dna-assembly]]></category>
            <category><![CDATA[plasmids]]></category>
            <category><![CDATA[molecular-biology]]></category>
            <dc:creator><![CDATA[Joshua Timmons]]></dc:creator>
            <pubDate>Mon, 13 Jan 2020 20:32:55 GMT</pubDate>
            <atom:updated>2020-01-13T20:32:55.481Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CJQxuHUhEbqxmyPH9oF7TQ.png" /></figure><p>Lattice’s goal is to build tools for biologists that make their jobs and research easier. One area where we do that is DNA assembly. Our collaborators and clients typically use Geneious, SnapGene, or Benchling to plan their plasmid designs, but when it comes to DNA assembly these programs do the bare minimum.</p><h3>Background</h3><p>The existing software tools’ approach to plasmid design is “fragment-based.” That means biologists specify their DNA fragments and assembly technique — Golden Gate, Gibson assembly®, etc. — and their software pieces those fragments together. For Golden Gate, that involves checking for valid and unique 4-bp overhangs. In Gibson assembly that involves validating homologous overhangs so fragments anneal to their neighbors. In the more advanced versions of these bio-design programs, like <a href="https://j5.jbei.org/j5manual/pages/1.html">j5</a>, PCR primers are created so fragments can assemble (after PCR amplification) while they otherwise would not.</p><p>The steps of fragment-based plasmid design are the bare minimum and are trivial to implement because they depend on the biologist doing most of the work. The biologist chooses the fragments and, unless it makes new primers (e.g. j5), the tool does nothing besides concatenate those fragments together.</p><p>The biggest shortcoming of fragment-based plasmid design is that it is backward. When creating a plasmid, it is the <em>plasmid</em> the biologist cares about, not the constituent fragments. Fragment-based plasmid design is akin to asking an architect to draft a list of building materials rather than a building. It puts the cart before the horse. An architect should be free to design their building and let software list out the building materials. Likewise, biologists should be free to create whatever plasmid sequence(s) they need for their project and let software tell them which fragments to build it with.</p><p>A second and more technical shortcoming of fragment-based plasmid design is that it produces sub-optimal plasmid designs. It assumes that the biologist has selected the best set of DNA fragments — by cost and likelihood to assemble. This assumption can easily fail.</p><p>For example, imagine a biologist creating a 12 kbp plasmid using a Gibson assembly of eight large fragments. Each of those eight fragments needs to be PCR’ed. But another plasmid covering 11 kbp of the plasmid’s sequence already exists in a remote DNA repository: Addgene. It would be better to re-use the 11 kbp sequence rather than PCR’ing each of the eight large fragments, but checking for all re-usable subsequences and finding the best combination of those subsequences is intractable for any plasmid or project of significant complexity. Addgene alone has 70,000 plasmids. Throw in DNA synthesis and the size of the plasmid’s design possibilities is innumerable, and the designer never realizes that there is a better design (an 11 kbp repository fragment + synthesis of the last kilobase pair). Fragment-based plasmid design will not help.</p><h3>REPP</h3><p>We made a free and open-source tool for plasmid design that addresses both noted shortcomings of fragment-based plasmid design. It creates cost-optimal Gibson assembly designs using repository-based DNA sequences as well as synthetic DNA. We call it “REPP,” an abbreviation of our preferred approach: “<strong>rep</strong>ository-based <strong>p</strong>lasmid design.”</p><p>Given a plasmid to design, REPP looks through DNA repositories/databases to find the least expensive combination of fragments to assemble it. We call this “inventory awareness.” Users can point to both local repositories and remote repositories. Local repositories can include the DNA available in labs’ fridges and freezers as well as the fridges and freezers of the labs’ neighbors. Remote repositories include <a href="http://parts.igem.org/Main_Page">iGEM</a>, <a href="http://www.addgene.org/">Addgene</a>, and <a href="https://dnasu.org/DNASU/Home.do">DNASU</a>.</p><p>REPP supports DNA synthesis and considers it for every plasmid design. REPP also avoids issues that can plague Gibson assembly like primers with ectopic binding sites, hairpins in the homology between fragments, and an excessive number of fragments.</p><p>REPP outputs a list of existing repository-based DNA fragments, primers to PCR amplify those fragments with homology for their neighbors, and synthetic fragments (if necessary).</p><h3>Example</h3><p>Here we provide an example of REPP in action. <a href="https://www.addgene.org/123158/">As0</a> is a plasmid containing a <a href="https://www.nature.com/articles/s41589-019-0244-3">bacterial arsenic sensor</a> and was created as part of a catalog of sensors created by <a href="https://www.addgene.org/Baojun_Wang/">the Wang lab</a> in 2019. We will create Gibson assembly design plans for As0 using REPP.</p><pre>repp make seq -i As0.fa -o As0.json --addgene --settings twist.yaml</pre><figure><img alt="An example of REPP designing the As0 plasmid with PCR fragment output." src="https://cdn-images-1.medium.com/max/614/1*atznu_xBsUfnY4_HyVMgMw.png" /></figure><p>With <a href="https://www.addgene.org/123158/sequences/">As0’s sequence</a> as input and Addgene as the sole source for DNA fragments, REPP creates a plasmid design using two PCR and two synthetic fragments. As0 has large regions of reused sequences from past plasmids from the Wang group. REPP recognizes that and recommended creating large PCR fragments from two of those source plasmids: <a href="https://www.addgene.org/78637/">#78637</a> and <a href="https://www.addgene.org/61439/">#61439</a>. It creates primers for both the PCR fragments and synthesizes regions of As0 that are not in Addgene (“2_synth” and “4_synth”).</p><p>The files and command we used to create As0’s design are on Github in the <a href="https://github.com/Lattice-Automation/repp/tree/master/examples">REPP’s examples directory</a>.</p><h3>Usage</h3><p>Users of REPP pass their plasmid’s sequence in either FASTA or GenBank format:</p><pre>repp make seq -i “target_seq.fa”</pre><p>Where “target_seq.fa” is a FASTA file that looks something like:</p><pre>&gt;2ndVal mScarlet-I<br>CAACCTTACCAGAGGGCGCCCCAG...</pre><p>REPP outputs a JSON file that looks something like the snippet below. The “solutions” property holds an array of solutions, each one comprising a number of “fragments” that are PCR fragments or synthetic fragments. Rarely, REPP will return a whole plasmid sequence if there is a match already within one of the source repositories.</p><pre>{<br>  &quot;target&quot;: &quot;2ndVal mScarlet-I&quot;,<br>  &quot;seq&quot;: &quot;CAACCTTACCAGAGGGCGCCCCAG...&quot;,<br>  &quot;solutions&quot;: [<br>    {<br>      &quot;count&quot;: 2,<br>      &quot;cost&quot;: 236.65,<br>      &quot;fragments&quot;: [<br>        {<br>          &quot;type&quot;: &quot;pcr&quot;,<br>          &quot;cost&quot;: 94.67,<br>          &quot;url&quot;: &quot;https://www.addgene.org/103998/&quot;,<br>          &quot;seq&quot;: &quot;ACAAATAAATGTCCAGACCTGCAG...&quot;,<br>          &quot;pcrSeq&quot;: &quot;ACAAATAAATGTCCAGACCTGCAG...&quot;,<br>          &quot;primers&quot;: [<br>            {<br>              &quot;seq&quot;: &quot;ACAAATAAATGTCCAGACCTGCA&quot;,<br>              &quot;strand&quot;: true,<br>              &quot;penalty&quot;: 4.406456,<br>              &quot;pairPenalty&quot;: 18.046225,<br>              &quot;tm&quot;: 58.594,<br>              &quot;gc&quot;: 39.13<br>            },<br>            {<br>              &quot;seq&quot;: &quot;CATATGTATATCTCCTTCTTAAATCT&quot;,<br>              &quot;strand&quot;: false,<br>              &quot;penalty&quot;: 13.639769,<br>              &quot;pairPenalty&quot;: 18.046225,<br>              &quot;tm&quot;: 52.36,<br>              &quot;gc&quot;: 26.923<br>            }<br>          ]<br>        },<br>        {<br>          &quot;id&quot;: &quot;103998-103998-synthesis-1&quot;,<br>          &quot;type&quot;: &quot;synthetic&quot;,<br>          &quot;cost&quot;: 129,<br>          &quot;seq&quot;: &quot;AGGAGATATACATATGGTGAGTAA...&quot;<br>        }<br>      ]<br>    }<br>  ]<br>}</pre><p><a href="https://lattice-automation.github.io/repp/repp_make_sequence.html">By default</a>, REPP checks for re-usable sequences among plasmids available from Addgene, iGEM, and DNASU. Users can specify additional, local repositories — for example, a fridge of DNA as a BLAST database — through the “-d” flag.</p><p>REPP also supports the standard fragment-based Gibson assembly:</p><pre>repp make fragments -i “multi_fragments.fa”</pre><p>As well as a feature-based assembly, where the input is a list of features to be pieced together in order:</p><pre>repp make features &quot;BBa_R0062 BBa_B0034 BBa_C0040 BBa_B0010 BBa_B0012&quot; --igem --backbone &quot;pSB1C3&quot; --enzymes &quot;PstI EcoRI&quot;</pre><h3>Installation</h3><p>REPP is available for Windows, macOS, and Linus via download from <a href="https://sourceforge.net/projects/repplasmid/files/">SourceForge</a>. The approach differs between MacOS/Linux and Windows:</p><h4><strong>macOS/Linux</strong></h4><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/275bf1af94ab71bcea17a3fee6c0c2e0/href">https://medium.com/media/275bf1af94ab71bcea17a3fee6c0c2e0/href</a></iframe><h4>Windows</h4><p>1. Download the latest repp_windows.*.zip from <a href="https://sourceforge.net/projects/repplasmid/files/">SourceForge</a></p><p>2. Unzip</p><p>3. Run repp_install.exe</p><h3><strong>Configuration</strong></h3><p>REPP is meant to be <a href="https://lattice-automation.github.io/repp/configuration">configurable</a> to the biologist’s design rules. Everything from the length of overhangs to the maximum allowable Tm of ectopic primer binding sites is adjustable. Importantly, biologists can adjust the cost-curves of DNA synthesis to match their provider — REPP supports both fixed bp pricing as a function of length (e.g. IDT’s step-wise cost curve) and variable, cost per bp pricing (e.g. Twist):</p><iframe src="" width="0" height="0" frameborder="0" scrolling="no"><a href="https://medium.com/media/62fdc1d5e8593735bafbbe191e49f076/href">https://medium.com/media/62fdc1d5e8593735bafbbe191e49f076/href</a></iframe><p>REPP installs a configuration file to $HOME/.repp/config.yaml for default, global settings but users can specify per-run settings through the <a href="https://lattice-automation.github.io/repp/repp_make_sequence.html">“ — settings” parameter</a>.</p><h3>More Information</h3><h4>Documentation</h4><p>Documentation for REPP can be found at <a href="https://lattice-automation.github.io/repp/">https://lattice-automation.github.io/repp/</a></p><h4>Manuscript</h4><p>A paper about REPP was published in and is available from <a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0223935">PLOS One</a>. In it, we discuss using REPP to build every iGEM part in the iGEM registry, as well as all plasmids submitted to Addgene in 2018. REPP reduced the cost of plasmid assembly by an estimated 34% versus assembly with synthetic fragments alone and 72% versus synthetic gene orders (two services offered by DNA synthesis companies). The larger the target plasmid sequence, the more likely REPP would find a cost-effective way to re-use DNA sequence from external repositories.</p><h4>Source Code</h4><p>The source code for REPP can be found on GitHub: <a href="https://github.com/Lattice-Automation/repp">https://github.com/Lattice-Automation/repp</a></p><h4>Feature/Modification Requests</h4><p>Please refer to the Issues page on GitHub, <a href="https://github.com/Lattice-Automation/repp">https://github.com/Lattice-Automation/repp</a>, or email us directly.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=312744dd484d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/repository-based-plasmid-design-312744dd484d">Repository-based plasmid design</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Visualize Your DNA Sequences with SeqViz]]></title>
            <link>https://medium.com/lattice-automation/visualize-your-dna-sequences-with-seqviz-b1d945eb9684?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/b1d945eb9684</guid>
            <category><![CDATA[reactjs]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[visualization]]></category>
            <category><![CDATA[synthetic-biology]]></category>
            <category><![CDATA[dna]]></category>
            <dc:creator><![CDATA[Lattice Automation]]></dc:creator>
            <pubDate>Wed, 14 Aug 2019 20:51:30 GMT</pubDate>
            <atom:updated>2020-02-03T21:29:58.455Z</atom:updated>
            <content:encoded><![CDATA[<h3>Visualize Your DNA with SeqViz</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fpL3ZFWR6o_jr3FU_0k3Lw.png" /></figure><p>The greatest thing about the internet is that it is a powerful conduit for sharing. And not just opinions and information — more and more tools, for tasks as generic as photo editing and text editing, to tasks as niche as making a meme gif or converting MPEG files to MP4, are being shared on the web for all to use. No license to purchase, no contracts to sign, just simple and ready to use.</p><p>Lattice Automation is releasing one such tool to add to your list: SeqViz, a simple, open-source, customizable sequence viewer for the web.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*V0S76GdfXYwAD9i93ifV5A.gif" /><figcaption>Demo at: <a href="https://tools.latticeautomation.com/seqviz/">https://tools.latticeautomation.com/seqviz/</a></figcaption></figure><h3>Simple to Use, Simple to Integrate</h3><h4>Install</h4><pre>npm install seqviz</pre><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*cEp8fLdfzLgVdgBvBkTBGw.png" /><figcaption>SeqViz provides circular and linear sequence viewers out of the box.</figcaption></figure><h4>Usage</h4><p>SeqViz is written in React but can in any Javascript environment.</p><h4>React</h4><pre>import { SeqViz } from &quot;seqviz&quot;;</pre><pre>export default () =&gt; (<br>  &lt;SeqViz<br>    name=&quot;J23100&quot;<br>    seq=&quot;TTGACGGCTAGCTCAGTCCTAGGTACAGTGCTAGC&quot;<br>    annotations={[{ name: &quot;promoter&quot;, start: 0, end: 34, direction: 1 }]}<br>  /&gt;<br>);</pre><h4>Vanilla-JS</h4><pre>&lt;script&gt;<br>  window.seqviz<br>    .Viewer(&quot;root&quot;, {<br>      name: &quot;L09136&quot;,<br>      seq: &quot;tcgcgcgtttcggtgatgacggtgaaaacctctgacacatgca&quot;<br>    })<br>    .render();<br>&lt;/script&gt;</pre><p>It also comes with parsers for common sequence file formats to facilitate ease of input. The library can currently read:</p><ul><li>FASTA format (.seq, .fa, .fas, .fasta)</li><li>GENBANK format (.gb, .gbk, .genbank, .ape)</li><li>SNAPGENE format (.dna)</li><li>JBEI format</li><li>SBOL 1.0 and SBOL 2.0 formats</li><li>Benchling format</li><li>BioBricks</li></ul><p>with more parser support coming soon. If the format you use is not yet supported by SeqViz, you can provide the library with a string or JSON representation of the sequence, or a string accession id, if the part can be found in the NCBI or iGEM registries. The library will fetch and interpret the data for you and render it to the sequence viewer. Read more on sequence data accepted by SeqViz in the <a href="https://github.com/Lattice-Automation/seqviz#Options">documentation</a>.</p><h3><strong>Open-sourced, Open to Improvement</strong></h3><blockquote>SeqViz is free and open-sourced for the betterment of the SynBio community</blockquote><p>You can see all of the code for the library as it continues to evolve for better performance, neater aesthetics, and more plentiful features. SeqViz is hosted and maintained by Lattice Automation but provided under the MIT Software License. This means that you are free to use, modify, and improve upon the library, while Lattice helps manage the source code, patch vulnerabilities, incrementally improve the tools, and even provide external utilities.</p><blockquote><em>SeqViz is built upon years of domain expertise in visualizing sequences.</em></blockquote><p>Lattice Automation has built custom sequence viewers for companies like Corteva Agriscience (see our <a href="https://latticeautomation.com/solutions">company engagements</a> for more information), and we also provide our own platform for end-to-end synthetic biology development in the form of <a href="https://loom.latticeautomation.com/">Loom Tools for Synthetic Biology</a> (engage with our Alpha testing program by messaging contact@latticeautomation.com). The SeqViz source code is a culmination of all that we have learned about displaying sequence information in the browser and it will continue to grow as we develop our commercial products.</p><p>We invite you to contribute feature requests and code additions directly to the library via GitHub. Visit <a href="https://github.com/Lattice-Automation/seqviz">https://github.com/Lattice-Automation/seqviz</a> for the full library and demonstration code, as well as information on how to start contributing.</p><h3><strong>Custom Look, Custom Feel</strong></h3><blockquote>A Google Maps for DNA Sequences.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EmhRzLjCJbTzV9ty0Hwk8A.png" /><figcaption>A demo of the SeqViz library used to visualize BioBricks is available at tools.latticeautomation.com/seqviz. The demo has user interface for most of the libraries customizable settings, including 1) how to display the sequence controlled by the Topology dropdown, 2) zooming in and out on the linear viewer controlled by the Zoom slider, 3) searching for subsequences in the search box, 4) controlling whether or not data is fetched from our annotation service with the Auto-annotate sliding checkbox, 5) controlling whether annotations, complement strand, and numbered axis are shown on the viewers with their respective sliding checkboxes, and 6) controlling whether cut sites of 4 enzymes specifically relevant to BioBricks are shown in the viewers with the Enzyme selection buttons</figcaption></figure><p>SeqViz was built to be a customizable library you can plug into an existing online resource (a website, wiki page, or webpage). While it provides you with an aesthetic viewer out of the box, it has no requirements on how you structure the user interface around it or how you pass it information, as long as the information is consumable by Javascript. If you have an existing ReactJS project, for example, you can simply import SeqViz as a dependency and use it as a React component. If you would like to allow people to interact with your sequence viewer you can build UI elements for</p><ol><li>Choosing whether to view the sequence in a Plasmid viewer or a Linear viewer or both</li><li>Choosing whether to show annotations</li><li>Choosing whether to show the complement strand nucleotide bases</li><li>Choosing whether to show an index line enumerating the bases</li><li>Zooming in and out on the Linear viewer</li><li>Choosing a list of colors to draw your annotations with</li><li>Searching sequences for a subsequence</li><li>Searching sequences for certain enzyme cut sites</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*et7zwQBOSi67eCHRk44cbQ.png" /><figcaption>The GenBank <strong>AF165912 </strong>sample part retrieved by accession id, displayed on a plasmid and linear sequence viewer with a <a href="https://github.com/Lattice-Automation/seqviz">search</a> for ATG start codon. You can generate the view above with the code below, passing in the appropriate id in the HTML input box.</figcaption></figure><h3><strong>Built for Interaction</strong></h3><blockquote>Click, drag, and scroll through your custom sequence viewer.</blockquote><p>SeqViz is built to be reactive and interactive. You can drag to select and click on annotations. As you interact with the viewer, the library will return an object containing information about your selected subsequence which contains helpful information like base-pair indices, length of the selection, Tm, GC%, name and type of feature selected, etc. Access this information through the onSelection() function to display it in your custom UI or use it in your application elsewhere. Read more about the onSelection() function on the <a href="https://github.com/Lattice-Automation/seqviz">documentation</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_ctj4yEaM2hw16Rl70NNNQ.png" /><figcaption>A demo of the SeqViz library used to visualize BioBricks is available at tools.latticeautomation.com/seqviz. If you select an annotation on the viewer you have the option of viewing its meta-information by clicking the Show Meta button.</figcaption></figure><p>The library will also return an object containing search results when the constructor is provided with a search query. This includes the full list of matches, including the start and end indices of each match. You can access this information through the onSearch() function (more information in the <a href="https://github.com/Lattice-Automation/seqviz#onsearch">repo documentation</a>).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ar5LXlz9B22UOyGCE4LKTA.png" /><figcaption>A demo of the SeqViz library used to visualize BioBricks is available at tools.latticeautomation.com/seqviz. If you search for a subsequence, matches will be highlighted on the linear sequence viewer. Matches are highlighted on both strands, in both forward and reverse directions.</figcaption></figure><h3><strong>Special Tooling</strong></h3><blockquote>SeqViz is open to custom integrations.</blockquote><p>The library already contains features tailored to iGEM.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RE-CoatrEa5ChhSXGjNLHQ.png" /><figcaption>A demo of the SeqViz library used to visualize BioBricks is available at tools.latticeautomation.com/seqviz. The demo is built to be a quick and easy start to exploring both the SeqViz library and the iGEM parts repository.</figcaption></figure><p>You can test out a demonstration of SeqViz for visualizing iGEM BioBricks at <a href="https://tools.latticeautomation.com/seqviz/">tools.latticeautomation.com/seqviz</a>.</p><p>For questions, concerns, and suggestions, feel free to reach out to us at contact@latticeautomation.com.</p><p><em>Written by: Raymond Chung</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b1d945eb9684" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/visualize-your-dna-sequences-with-seqviz-b1d945eb9684">Visualize Your DNA Sequences with SeqViz</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Combinatorial MoClo Assembly]]></title>
            <link>https://medium.com/lattice-automation/combinatorial-moclo-assembly-94b766b7c1b3?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/94b766b7c1b3</guid>
            <category><![CDATA[dna]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[dna-assembly]]></category>
            <category><![CDATA[computational-biology]]></category>
            <category><![CDATA[synthetic-biology]]></category>
            <dc:creator><![CDATA[Joshua Timmons]]></dc:creator>
            <pubDate>Wed, 22 May 2019 18:35:53 GMT</pubDate>
            <atom:updated>2019-05-22T18:35:53.033Z</atom:updated>
            <content:encoded><![CDATA[<iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FnKIRiPCeOBw%3Ffeature%3Doembed&amp;url=http%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DnKIRiPCeOBw&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FnKIRiPCeOBw%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/068716a28d6a1ef284ad37ec4e0ad5f2/href">https://medium.com/media/068716a28d6a1ef284ad37ec4e0ad5f2/href</a></iframe><p>Lattice’s MoClo assembly tool was one of the first things I worked on at Lattice. Its goal was ease of use with large combinatorial output — two things that normally conflict.</p><p>The tool had a scientist-friendly interface for generating combinatorial designs with for MoClo assembly. It output plate maps, liquid handling instructions, and the fully assembled &amp; annotated plasmid maps.</p><p>MoClo itself is a variant of the Golden Gate DNA assembly method, where BsaI and BpiI restriction enzyme sites flank specific DNA sequences (parts) in a standardized part library ¹. This method, and others like it, add standardization and interoperability of parts in designs at the cost of scars (the short, extra DNA sequences that remain after enzyme cleavage and ligation) ².</p><p>Researchers using MoClo can generate large libraries of plasmids, circular pieces of DNA, with many different combinations of these parts, tuned to their target organism. However, the number of possible combinations of parts increases exponentially with the number of parts that a researcher wants to test. This can easily lead to design spaces that are too large for researchers to tackle manually in the lab.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*ba3wSXmc4sBwJJK9X6tePA.jpeg" /><figcaption>(1) Researchers first upload Genbank files for all DNA parts &amp; destination vectors they wish to use. (2) Next, DNA parts to be included in the assembly are selected. (3) The tool will then generate a picklist for an automated liquid handler, as well as plate maps to assist with manual population with DNA parts and enzyme mastermix. (4) Using the DNA &amp; reagents plates, as well as the generated picklist, researchers execute the assembly of the cloning reactions on the liquid handler. (5) Once complete, the reactions are transformed and plated for downstream analysis.</figcaption></figure><p>To make this sort of large-scale combinatorial design accessible to many researchers, we built a user-centric web-based MoClo assembly tool. From it, researchers could assemble and inspect the results of combinatorial MoClo in a single, easy-to-use interface. Researchers uploaded their standard bio files — Genbank, Fasta, etc. — and the tool parses them into basic DNA parts. It did so using a simulated digest of BsaI and BpiI, looking for, and saving, all the post-digestion fragments as DNA parts. When parsing the files, we inferred part types from file annotations and binned parts by their type in a dropdown navigator to the right of the “workspace.”</p><p>From the navigator, researchers would drag single parts, or groups of parts, into the workspace. Researchers could then select either a single part, like a promoter, or dozens, placing all of them into a single group in the workspace.</p><p>Researchers could also apply unique filters to each group of parts. For example, ignoring all parts in a group with a GC% greater than 75%.</p><p>After creating groups of parts — in the locations of a plasmid where they would like them in the output — the researcher would click “Assemble.” The MoClo Assembly tool would then find all combinations of parts that would be compatible to generate a circularized plasmid.</p><p>We built in several other views so researchers could inspect the output. The first was a plate viewer where each well in all design plates was inspectable — showing where the MoClo Assembly tool had allocated fragments. The plate viewer also acted as an essential guide for setting up the assembly and was exportable as a PDF for the biologists to use in the lab. Another view was at the project-level, where the user could store past designs and compare them to the most recent run of the assembly tool.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/700/1*oZaQXoVtFZ_7DvRQz3WsMA.jpeg" /><figcaption>Reaction setup time for 96 reactions assembled manually, via a liquid handler, and by an acoustic dispenser. Manual assembly took 2 h 10 min, all of which was hands-on time. The liquid handler took a similar amount of time (2 h 6 min) to execute the pipetting commands, however, only a fraction of that time (5 min) was hands-on. The acoustic dispenser took significantly less time to execute liquid transfers (5 min) and took minimal hands-on time (5 min).</figcaption></figure><p>Our work culminated in the publication of an article in the Journal of Visual Experiments (JoVE), where Luis Ortiz, a biologist, automated the design and construction of 96 composite DNA variants using our tool. By Luis’ estimates, the tool saved him 95% of his typical hands-on time when combined with a lab robot compared to executing the same number of assemblies manually.</p><p>The tool was a small and early example of the kinds of processes we automate at Lattice Automation. Customers come to us with their own DNA assembly challenges and work with us to develop a solution that fits their needs and connects to external services like LIMS, liquid handling robotics, and electronic lab notebooks. If you are interested in using this tool or developing something similar, please contact us.</p><p>[1]: <a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0016765">https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0016765</a></p><p>[2]: <a href="https://pubs.acs.org/doi/abs/10.1021/acssynbio.5b00124?src=recsys&amp;journalCode=asbcd6">https://pubs.acs.org/doi/abs/10.1021/acssynbio.5b00124?src=recsys&amp;journalCode=asbcd6</a></p><p>[3]: <a href="https://www.jove.com/video/54703/automated-robotic-liquid-handling-assembly-of-modular-dna-devices">https://www.jove.com/video/54703/automated-robotic-liquid-handling-assembly-of-modular-dna-devices</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=94b766b7c1b3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/combinatorial-moclo-assembly-94b766b7c1b3">Combinatorial MoClo Assembly</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Technology Behind Lattice Automation]]></title>
            <link>https://medium.com/lattice-automation/technology-behind-lattice-automation-6df24cd150c6?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/6df24cd150c6</guid>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[dna-assembly]]></category>
            <category><![CDATA[biodesign]]></category>
            <category><![CDATA[synthetic-biology]]></category>
            <category><![CDATA[algorithms]]></category>
            <dc:creator><![CDATA[Lattice Automation]]></dc:creator>
            <pubDate>Thu, 25 Apr 2019 13:17:19 GMT</pubDate>
            <atom:updated>2019-04-24T17:22:55.306Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://en.wikipedia.org/wiki/Alberto_Sangiovanni-Vincentelli">Alberto Sangiovanni-Vincentelli</a>. Quite a mouthful. Particularly for someone from Michigan. Good thing he let me just call him “Alberto”. Alberto was my advisor at UC Berkeley. He was (is) at the forefront of <a href="https://www.eetimes.com/document.asp?doc_id=1204965#">“Platform Based Design” </a>which is an approach to the design of embedded electronic systems. Lattice was started with this philosophy and in that sense is very uniquely positioned to think about problems in a very modular, formalized way. Lattice has very strong ties to academic research (I am a professor at Boston University at the end of the day). <strong>Research is in Lattice’s DNA.</strong> I probably should have called the company Lattice Research instead of Lattice Automation.</p><p>I have a lot of people ask me about our technology and I inevitably have to forward them a bunch of papers. <strong>This post is my way of capturing the top 10 papers from my research that represent what Lattice is all about.</strong> Without further ado…</p><p><strong>Honorable Mention</strong> — <a href="https://www.jove.com/video/54703/automated-robotic-liquid-handling-assembly-of-modular-dna-devices"><em>Automated robotic liquid handling assembly of modular DNA devices</em></a><em>. </em>JOVE, 2017</p><p>This paper actually has two Lattice authors (Joshua Timmons and Lloyd McCarthy). A good overview of what it takes to get liquid handling robotic automation to work in a very practical manner and presents a very prototype design application built to user specifications.<strong> Good example of a computational and experimental collaboration between organizations.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bfLqfzIKmIA_whPD8FzVyg.png" /></figure><p><strong>10. </strong><a href="https://www.cell.com/trends/biotechnology/fulltext/S0167-7799(13)00223-0"><em>Bio-design automation: software+ biology+ robots</em></a><em>. </em>Trends in Biotechnology, 2014</p><p>A position paper. Discusses my philosophy on bio-design automation. Explicitly exposes my view that <strong>“Specification” is an equal part</strong> of the design-build-test cycle. This separation is a key part of Platform-based Design. Credit to Swapnil Bhatia for the catchy title.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/410/1*dyhm7ciN0y50WVIPzUSJ9A.png" /></figure><p><strong>9.</strong> <a href="https://www.sciencedirect.com/science/article/pii/B978012385120800005X"><em>Developer’s and User’s Guide to Clotho v2. 0: A Software Platform for the Creation of Synthetic Biological Systems</em></a><em>. </em>Methods in Enzymology, 2011</p><p>In general Clotho was poorly published (sigh…). This continues to be a challenge for me with large scale software engineering efforts (hence the birth of Lattice). This is not a great paper and is already VERY dated but it gives you some idea of the concepts Clotho was trying to establish. The most important part is that it <strong>exposes the data types present in Clotho’s data model </strong>and how they work together. You can see data elements that SBOL has incorporated as well. Having this data model explicitly exposed was Clotho’s key contribution to the field.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/751/1*9mRaf05XvGZZJJmDAEjshQ.png" /></figure><p><strong>8.</strong> <a href="https://jbioleng.biomedcentral.com/articles/10.1186/1754-1611-6-1"><em>DeviceEditor visual biological CAD canvas</em></a><em>. </em>Journal of Biological Engineering, 2012</p><p>In 2009 the UC Berkeley Software iGEM team created a tool for Clotho called <a href="http://2009.igem.org/Team:Berkeley_Software/Spectacles">“Spectacles”</a>. This was a drag-and-drop environment to design composite DNA part based designs. This has been done “better” now by programs like <a href="https://github.com/SynBioDex/SBOLDesigner">SBOL Designer</a>. This work also seeded some of the tools created by <a href="https://teselagen.com/">Teselagen</a>. <strong>It is unique in that it has a visual canvas that can be controlled by a domain specific language (Eugene). </strong>I personally do not feel that visual environments like this are the future. I don’t think humans visually designing things is going to 1) scale and 2) be semantically rich enough to specify designs. However this does provide a small window into how this style of approach got started.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/785/0*4IU7xJKonUSZi73m.jpg" /></figure><p><strong>7.</strong> <a href="https://pubs.acs.org/doi/abs/10.1021/acssynbio.5b00219"><em>Merlin: Computer-aided oligonucleotide design for large scale genome engineering with MAGE</em></a><em>. </em>ACS Synthetic Biology, 2016</p><p>Merlin is a tool to design oligos for <a href="https://www.nature.com/articles/nature08187">Mage (a Church Lab technology)</a>. This paper is another demonstration of a collaboration between groups and the <strong>formalization of a problem</strong> that had a very naive original implementation. I would have liked to follow this up with more experimental work. This remains an open problem and one that several of our customers have been interested in exploring.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*QPMExQtCwmxx20NW.gif" /></figure><p><strong>6.</strong> <a href="https://pubs.acs.org/doi/abs/10.1021/acssynbio.5b00232"><em>Double dutch: A tool for designing combinatorial libraries of biological systems</em></a><em>. </em>ACS Synthetic Biology, 2016</p><p>An example design of experiments based approach. This work has a nice open <a href="https://github.com/CIDARLAB/doubledutch">source set of tools</a> to go with it. The paper does a decent job of also <strong>outlining the algorithms</strong> in the main body of the paper and presenting the results over a couple of case studies. Really nice work lead by <a href="https://scholar.google.com/citations?user=rxAi6qQAAAAJ&amp;hl=en">Nic Roehner</a>. The name is a play on words since the the statistical software <a href="https://www.jmp.com/en_us/home.html">JMP</a> was used (i.e. jump rope).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/679/1*oWpssLZC6YK8sIQ_gec8fg.png" /></figure><p><strong>5.</strong> <a href="https://pubs.acs.org/doi/abs/10.1021/sb300030d"><em>An end-to-end workflow for engineering of biological networks from high-level specifications</em></a><em>. </em>ACS Synthetic Biology 2012</p><p>This work is often referred to as the “TASBE project” and was joint work done with BBN Technologies and MIT as part of DARPA work that preceded the 1000 Molecules and <a href="https://www.darpa.mil/program/living-foundries">Living Foundries</a> programs. Lead by <a href="https://scholar.google.com/citations?user=_ZSyxZ8AAAAJ&amp;hl=en">Jake Beal</a> at BBN,<strong> this was one of the first end-to-end tool-chains demonstrated for synthetic biology. </strong>Most of the software has been replaced with more sophisticated approaches but it was a nice holistic approach for its time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/913/1*hIS8X9AsKx_oDMW8PaRStQ.png" /></figure><p><strong>4.</strong> <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2860133/"><em>Algorithms for automated DNA assembly.</em></a><em> </em>Nucleic Acids Research, 2010</p><p>Elegant dynamic programming formulation for binary assembly of DNA parts. <a href="https://scholar.google.com/citations?user=GJL39o8AAAAJ&amp;hl=en">Evan Appleton’s </a>follow up paper (“ <a href="https://www.nature.com/nmeth/journal/v11/n6/abs/nmeth.2939.html">Interactive assembly algorithms for molecular cloning</a>”) is likely more complete and has a software tool called “Raven” to back it up. Also this work was surpassed in some regards by <a href="https://pubs.acs.org/doi/abs/10.1021/sb400161v">“Heuristic for Maximizing DNA Reuse in Synthetic DNA Library Assembly”</a>. However, <strong>this paper laid the foundation and likely is superior in its simplicity and clarity. </strong>This<strong> </strong>also is my last real first author technical paper so I have a soft spot for it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hIJAbbHnJoVybT5AG2WZiA.png" /></figure><p><strong>3.</strong> <a href="https://pubs.acs.org/doi/abs/10.1021/acssynbio.7b00154"><em>Genetic Design via Combinatorial Constraint Specification</em></a><em>. </em>ACS Synthetic Biology, 2017</p><p><strong>This paper is criminally under recognized in my opinion</strong>. A truly elegant solution to representing genetic designs by <a href="https://scholar.google.com/citations?user=iKsqYU4AAAAJ&amp;hl=en">Swapnil Bhatia</a>. The formalism lets you both generate and verify designs. It also is amenable to numerous formal operations that are correct-by-construction. Storage in graph based databases is also natural. <strong>This type of approach is the future of design specification</strong>. Lattice tools all will support this data model and its associated operations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/572/1*EGVqOu0am8GJglQ-JJ8XUA.png" /></figure><p><strong>2.</strong> <a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0018882"><em>Eugene–a domain specific language for specifying and constraining synthetic biological parts, devices, and systems</em></a><em>.</em> PLoS One, 2011</p><p>Lots of great concepts here. The idea of a programming language that allows you to specify genetic constraints while providing the control flow statements that allow you to make decisions on those constraints is very powerful. <strong>This work is still ahead of its time. </strong><a href="https://scholar.google.com/citations?user=jviyfgcAAAAJ&amp;hl=en">Ernst Oberortner</a> built other publications on top of this which are more formal but this is the original work. The funny thing is that as one of my most cited papers, I often get international students reaching out to me as Dr. Bilitchenko.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m6gvXlPJRDYAULMmg7ExIg.png" /></figure><ol><li><a href="http://science.sciencemag.org/content/352/6281/aac7341"><em>Genetic circuit design automation</em></a><em>.</em> Science, 2016</li></ol><p><strong>A paper that truly puts it all together</strong>. Extremely rigorous experimental work, cutting edge computational work, true inter-university collaboration, and inter-disciplinary expertise. All of this is tied together by an air tight vision of how parts based synthetic biology should be done. This is the paper Chris Voigt and I envisioned when we first met in 2007. This work serves as the foundation as well for the creation of <a href="https://www.asimov.io/">Asimov</a>. The <a href="http://cellocad.org/">Cello</a> design software represents a true separation of specification and design.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6btHDq8Am23rPO8WOFbtow.jpeg" /></figure><p><strong>Douglas Densmore</strong></p><p><strong>President</strong></p><p><strong>doug@latticeautomation.com</strong></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6df24cd150c6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/technology-behind-lattice-automation-6df24cd150c6">Technology Behind Lattice Automation</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[History of Lattice Automation]]></title>
            <link>https://medium.com/lattice-automation/history-of-lattice-automation-656e5f6812f3?source=rss----736e74318f86---4</link>
            <guid isPermaLink="false">https://medium.com/p/656e5f6812f3</guid>
            <category><![CDATA[synthetic-biology]]></category>
            <category><![CDATA[entrepreneurship]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[biodesign]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Lattice Automation]]></dc:creator>
            <pubDate>Thu, 25 Apr 2019 13:17:02 GMT</pubDate>
            <atom:updated>2019-05-24T15:17:38.228Z</atom:updated>
            <content:encoded><![CDATA[<p>“We use spreadsheets”…</p><p>This was the response from <a href="http://web.mit.edu/voigtlab/">Chris Voigt</a> in 2007 when I asked him what software he used to keep track of the hundreds of DNA sequences for his lab’s genetic circuits. The same software used by accountants was being used to design cutting edge synthetic biological designs. It was being used to not only store the DNA sequence of the designs but then also perform computations on those sequences. It struck me that this system was incredibly fragile and used Excel in a way that it never was intended. Chris’ lab was (is) extremely successful. They weren’t doing this out of “laziness” or ignorance. Ostensibly this was used for two reasons. The first is that biologists could quickly learn Excel. The second was the lack of dedicated “bio-design automation” software tools to engineer biology. I knew instantly that there was a place for me in Synthetic Biology.</p><p>After this conversation, I set out on what has now been a twelve-year process to improve on both points. The first point is to make software that biologists want and is familiar to them. The second is to make bio-design automation a first-class research citizen. Along the way, we should discover new approaches that biologists never would have thought possible or have thought to ask for. We should expand biologists’ “comfort zone” regarding computing and work together to formalize problems as algorithms so that more computer scientists can get involved. I think I have been successful in many aspects but much work remains. <a href="https://www.latticeautomation.com/"><strong>Lattice</strong></a><strong> represents the disciplined, commercial approach to this iterative process between biological experts and computer scientists.</strong></p><h3>In the beginning…</h3><p>My PhD is from UC Berkeley in Electrical Engineering. Much of my work was involved in creating two tools: <a href="https://ptolemy.berkeley.edu/projects/embedded/metropolis/">Metropolis and Metro II</a>. These are embedded electronic design frameworks that focus on the concept of <a href="https://en.wikipedia.org/wiki/Separation_of_concerns">“separation of concerns”</a>. In particular separating out the three C’s: computation, communication, and coordination. These tools allowed you to create <a href="https://ieeexplore.ieee.org/document/736561">models of computation </a>via distinct software elements including processes (computation), media (communication), and schedulers (coordination). The system simulation was controlled by <a href="https://ptolemy.berkeley.edu/papers/96/denotational/">“tagged events”</a> which not only modeled the causality and results of computation but also allowed “quantity managers” to annotate these events with values that provided the cost of the operations (e.g. power, time). This entire research effort was part of “System Level Design” and was (is) at the cutting edge of <a href="https://en.wikipedia.org/wiki/Electronic_design_automation">Electronic Design Automation</a> (EDA).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/320/0*q3qCbENASTYyPpCS.jpg" /><figcaption><strong>Platform-based Design. </strong>Applications are constrained to match a platform’s capabilities (behaviors are restricted; the top funnel) and then the platform is mapped (functionally assigned to physical resources; bottom funnel) to one of many implementations. This is is a core concept in system level design of embedded systems.</figcaption></figure><p>Sounds cool right? Actually, it is a bit complicated. The issue right now in EDA is that the lower levels of the design chain (e.g. physical design and logic synthesis) are well understood and mathematically rigorous (i.e. algorithmically hard). Therefore researchers are moving further up the design chain into more and more abstract design approaches, domain specific languages, and various verification frameworks. Many of these approaches are not immediately needed and the “abstraction gap” is not well bridged, leading to many iterations on designs and models that don’t represent well what the final implementation is doing. I ultimately was not convinced that I was the right person to solve these problems. I also wasn’t convinced that there was an immediate need to solve these problems. However, I was pretty sure I could beat a spreadsheet.</p><p>After I graduated, I decided to stay at UC Berkeley to do a post doc. I had a UC Chancellor’s Postdoctoral Fellowship which meant I did not have to work on a grant-based project. I could do whatever I wanted provided that it promoted the ideas of our research group (i.e. Platform-based Design). I figured I would try to apply “separation of concerns” to a toolset for synthetic biology. Borrowing a DNA “parts” based approach from Voigt, the idea was that the function you wanted a biological system to do (“the what”) would be constrained to a collection of viable parts. Those parts would be the platform. This would then be “mapped” to a final, linear sequence of DNA along with a host organism (“the how”). Environmental context, performance requirements, and costs would all form the basis for constraints (i.e. the funnels). I thought about how these ideas might apply to Platform-based Design and made some rough sketches of what might be the platform and what might be the architecture. <strong>This was groundbreaking in 2007.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*e5_1FbIC6ndkT_TPTqMmOw.png" /><figcaption><strong>Platform-based Design for Synthetic Biology (circa 2007).</strong> This was wildly novel but very exciting to Synthetic Biologists in the Bay Area including Chris Voigt, Chris Anderson, Adam Arkin, and Jay Keasling.</figcaption></figure><h3>Clotho emerges</h3><p>The first step was to get rid of Excel with a database (mySQL) and then build a layer of simple tools on top of this (Java for the backend and Java Swing for the UI). My first tools where a simple database browser, DNA sequence editor, and lightweight scripting interface to perform simple operations on the sequence. The tools were purposely made very modular since I had NO clue what tools to make and which ones biologists would want. I started out from the beginning to build a tool that allowed new “apps” to be added and open source so that I would NOT have to be the one to write all the apps.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8hhESXKKrT5t2h7J6Qz1aA.png" /><figcaption><strong>Clotho 0.0 Architecture and UI.</strong> A very object-oriented approach was taken in the first design of Clotho to allow new objects to be added easily by experts. The design explicitly adhered to separation of concerns by creating clear “algorithms”, “functions”, “views”, “connectors” and “interfaces”.</figcaption></figure><p>I needed a name for this tool suite. I didn’t want it to sound bio-ish and I also didn’t want it to imply a single function (there were more functions to come). Also, I hated <a href="https://www.youtube.com/watch?v=9STeegpxSb0">hybrid names</a>. Inspired by the auspicious sounding name of <a href="https://ptolemy.berkeley.edu/~eal/">Edward Lee’s</a> <a href="https://ptolemy.berkeley.edu/">Ptolemy</a> project, I picked the name “Clotho”. <strong>Clotho was the Greek fate that spun the thread of life. Sounded appropriate.</strong></p><h3>iGEM, Chris Anderson, and JBEI</h3><p>It quickly became clear that going back and forth from UC Berkeley to UCSF was not going to be manageable long term. It is amazing how much of a barrier a simple bridge is. Chris Voigt pointed me to a former student of his named Chris Anderson at UC Berkeley. Anderson was a new professor and very interested in this vision of software tools for synthetic biology. He quickly suggested that I lead a complementary software effort to the traditional wetlab team at UC Berkeley for the <a href="https://igem.org/Main_Page">International Genetically Engineered Machine</a> (iGEM) competition. To seed the project, we would use Clotho.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/0*rdhuZfAJfPitN9jL.jpg" /><figcaption><strong>Clotho and her “friends” </strong><a href="https://eugene.cidarlab.org/"><strong>Eugene</strong></a><strong> and Kelper. </strong>2009’s UC Berkeley iGEM team added a domain specific language (Eugene) and a workflow description environment (Kepler) to the Clotho ecosystem of tools. This also included a visual design environment called “Spectacles” which was a precursor to many visual, parts-based synthetic biology tools.</figcaption></figure><p>The <a href="http://2008.igem.org/Team:UC_Berkeley_Tools">2008 iGEM Team</a> ended up winning the inaugural software competition at iGEM. Tools were expanded in <a href="http://2009.igem.org/Team:Berkeley_Software">2009</a> (e.g. <a href="https://eugene.cidarlab.org/">Eugene</a>) and we won the software competition again. This was a great way to not only develop software that was needed but also get feedback from the larger community. It became clear that there was a hunger for this technology in a way that I was not experiencing with my EDA tools. <strong>There was also a wealth of “low hanging fruit” in the form of unformulated problems that if handed over to trained computer scientists, novel solutions would emerge.</strong></p><p>Taking my initial post doc experience I went on the job market as someone that would both do embedded systems research as well as synthetic biology design automation. That didn’t work. My message was muddled and didn’t make it clear how I would run a group, get external funding, or clearly contribute to either community. Furthermore, my contributions to embedded system design were likely not strong enough to be a faculty member. I knew I wanted to give academia another chance. I also knew that my contributions to synthetic biology (while early) were very novel and put me into a research area few were exploring. So I decided to focus on just bio-design automation and do so in a more industrial setting. I did so via another post doc at the <a href="https://www.jbei.org/">Joint Bio-Energy Institute</a> working for <a href="https://www.pbs.org/wgbh/nova/video/profile-jay-keasling">Jay Keasling</a> under the supervision of <a href="http://biosciences.lbl.gov/profiles/nathan-hillson/">Nathan Hillson</a>. There I helped to create <a href="https://jbioleng.biomedcentral.com/articles/10.1186/1754-1611-6-1">Device Editor</a>. This was a design canvas that allowed users to create designs for combinatorial genetic design. Aspects of this software have gone on to live in commercial forms via Nathan’s company <a href="https://teselagen.com/">Teselagen</a>. In addition, Chris Anderson and I continued to work on new versions of Clotho.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PsYbNJsg-Iz_PeltiuJ1ow.png" /><figcaption><strong>Clotho 1.0 Architecture and UI.</strong> Clotho continued to evolve. Java’s Hibernate was used to manage storage and additional database connections were added. More UI development increased the number of tools available and introduced innovations like visual part based design, and modular app based plug-ins</figcaption></figure><h3>Algorithms for DNA assembly</h3><p>As my job search had demonstrated, simply making some software tools was not going to be enough for me to land an academic job. <strong>I needed to clearly demonstrate an academic innovation and identity in synthetic biology. </strong>At the time Chris Anderson had a Beckman Coulter liquid handling robot in his lab. It was not fully utilized since most of the DNA assemblies that were being done were “one-offs”. No one was looking at all the assemblies that needed to be done, scheduling them, and determining which aspects could be shared across assemblies. This seemed like the perfect problem to solve and one that would work well in the Clotho framework given its access to a database of parts and designs of multiple users.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/305/1*BHSgg4BXiPp63Cd_kBlePg.png" /><figcaption><strong>DNA assembly formulated as a Computer Science problem. </strong>This work showed that biological automation challenges in synthetic biology could be put into elegant computer science frameworks given the right mix of biological and computer science teams and experiences.</figcaption></figure><p>It just so happened that at that time <a href="http://www.csl.cornell.edu/~cbatten/">Chris Batten</a> was doing a post doc at UC Berkeley. Chris had been a student at MIT and had been involved in iGEM during his time there (despite being a computer architecture graduate student). He had given some thought to formalizing <a href="http://parts.igem.org/Help:An_Introduction_to_BioBricks">“binary assembly”</a> style DNA construction as a series of “assembly graphs”. He, however, had not finished this work, created a software framework for it, or worked with actual biologists to vet it. <strong>Together along with Anderson lab students, we took on this challenge. </strong>The end result was “<a href="https://www.ncbi.nlm.nih.gov/pubmed/20335162">Algorithms for automated DNA assembly</a>” published in Nucleic Acids Research in 2010 and clearly formulated the problem to be solved. It represented a clear design automation challenge being addressed in an interdisciplinary setting. It utilized solid computer science approaches to a problem encoding biological realities. It was very well received. Using this, <a href="https://eugene.cidarlab.org/">the Eugene programming language</a>, and new Clotho tools, I was able to successfully secure a position at Boston University. <strong>I had officially given up EDA for BDA.</strong></p><h3>Funding (and publishing) Bio-Design Automation</h3><p>The first task of any new junior professor is to get a grant. A major one for engineers is the <a href="https://www.nsf.gov/pubs/2017/nsf17537/nsf17537.htm">NSF CAREER Award</a>. My award was entitled <a href="https://www.nsf.gov/awardsearch/showAward?AWD_ID=1253856">“Design Automation Infrastructure for DNA Assembly in Synthetic Biology”</a>. This allowed me to create software to extend my automated DNA assembly work. In addition, Chris Anderson and I received an NSF Advances in Biological Informatics award entitled <a href="https://www.nsf.gov/awardsearch/showAward?AWD_ID=1147158">“A Modular, Community Based Design Platform for Synthetic Biology (Clotho)”</a>. These two awards formed the foundation of my funding early on in my career.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oqEIMPyraALGpDcC2J7xDA.png" /><figcaption><strong>NSF Funded Bio-Design Automation Efforts.</strong> The National Science Foundation (NSF) funded early work for a new version of Clotho (left) as well as a software “stack” that included mechanisms to automate the assembly of DNA (called Puppeteer).</figcaption></figure><p>Something that quickly became clear however was that funding from sources like DoD or NIH was going to require strong biological applications. I have always been “biologically agnostic”. The software has purposefully been that way as well in order to appeal to the largest possible audience. Developing bio-design automation software without strong application ties quickly became a funding nightmare. Furthermore, publishing increasingly required highlighting the application of this work (particularly in journals like Nature, Science, or Nature Biotechnology) relegating the algorithms to the supplemental information. I determined that it simply was not practical to develop software like this in academia. It did not yield funding (to run the lab) and it did not develop into highly visible publications (to promote the lab and the students). I addressed the open source community needs via the creation of the <a href="https://www.nonasoftware.org/">Nona Research Foundation</a>. However, there was a strong pull from commercial partners to develop the software. <strong>It became clear that an external commercial organization was needed.</strong></p><h3>Lattice Automation is born</h3><p>A professor really is only as good as his or her students. They not only do much of the grunt work but good students also introduce new ideas. I was fortunate to have a great first post doc. <a href="https://scholar.google.com/citations?user=iKsqYU4AAAAJ&amp;hl=en">Swapnil Bhatia</a> basically joined my group as I drove from Berkeley to Boston. I literally interviewed him on the phone on my drive out. Swapnil has the amazing ability to quickly synthesize a problem, abstract it into an algorithm, and then apply the appropriate domain specific heuristics. He is also meticulous in his approach and rigorous in his analysis. He presents ideas clearly and knows how to promote them.</p><p>In 2013, I knew that the software development work that I wanted to do, couldn’t happen in academia. I also knew the companies wanted my tools. I also knew, however, that a strong research foundation should be at the core of the company. I knew that we did NOT know what tools would be needed in 2023. <strong>We needed to create a company with a research culture that would work with customers (as I had with academics) to create customized solutions to hard problems.</strong></p><p>So, together with Swapnil, we decided to start Lattice. We knew we were going to have to involve BU so we pursued (and were awarded) an <a href="http://blogs.bu.edu/otd/2013/04/11/high-tech-50000-grant-winners/">Ignition Award </a>from BU. We also did <a href="https://www.nsf.gov/awardsearch/showAward?AWD_ID=1355868">NSF’s iCorp</a> to develop our business model. We set up shop in the BU Photonics Center’s Business Incubator to be close to my academic lab and stay in the vibrant community created by the campus atmosphere. I did not pursue additional venture funding since frankly we 1) did not need it for a pure software business and 2) the “pitch” to receive money to really “move the needle” was going to be complete bullshit. Instead of spending time with investors I started to reach out to all of the companies that had expressed interest previously. I put my efforts not into pitch decks but ideas for software and developing relationships with customers.</p><p>The approach worked. We suddenly realized that we were going to need to start to staff Lattice in earnest since both of us (still in academic positions) were not going to be sufficient. <a href="https://www.linkedin.com/in/kevinleshane/">Kevin LeShane</a> was a computer engineering MS student that was working for me on an embedded sensors project. Kevin was extremely reliable, a talented programmer, and possessed a high level of attention to detail. He also had zero ego regarding helping out on the numerous tasks needed to run a company (e.g. taxes, payroll, paperwork, etc.). I knew that Kevin should be “employee number 1”. With Kevin on-board we were able to write successfully NSF SBIR Phase 1 ( “<a href="https://www.nsf.gov/awardsearch/showAward?AWD_ID=1519998">Commercialization of Bio-Design Automation Software for Synthetic Biology</a>”) and Phase 2 (“ <a href="https://www.nsf.gov/awardsearch/showAward?AWD_ID=1660188&amp;HistoricalAwards=false">Fully Integrated Design and Automated Fabrication Services Software Platform for Engineering Living Systems</a>”) grants. <strong>Lattice Automation was officially open for business. </strong>Since then we have received over $2 million in grants and customer contracts, hired tremendously talented folks (for example software engineer <a href="https://www.linkedin.com/in/jjtimmons/">Joshua Timmons</a>), doubled in size, and even put out some <a href="https://www.jove.com/video/54703/automated-robotic-liquid-handling-assembly-of-modular-dna-devices">academic work</a>. The future is bright and I am very excited about 2019 and beyond!</p><h3>What’s Next?</h3><p>Today Lattice is focusing on tough problems facing bio-design automation in Synthetic Biology. In particular:</p><ol><li><strong>Formal representations of design. </strong>You probably laughed at the use of spreadsheets but what about GenBank files? Why are these any better? They are not built for the needs of 21st century bio-design. We need more standards like <a href="http://sbolstandard.org/">SBOL</a> to help pave the way for more formalized (i.e. well defined, rigorous, and analyzable) data exchange standards. We also need data structures the <a href="https://pubs.acs.org/doi/abs/10.1021/acssynbio.7b00154">represent combinatorial design</a>, DNA assembly, and <a href="https://jbioleng.biomedcentral.com/articles/10.1186/1754-1611-4-13">experimental protocols</a>. Lattice is working on all these problems and deploying them to our customers.</li><li><strong>Design of experiments. </strong>The “design space” in synthetic biology is very large. Design has to be about not only “taking more shots on goal” but making those shots be smarter. Every design someone makes needs to help to answer more than just “does this work”. It needs to explore the design space so that subsequent or derivative designs can be made more quickly, inexpensively, and higher performing.</li><li><strong>Hardware-software-wetware co-design.</strong> We are <a href="http://science.sciencemag.org/content/352/6281/aac7341">exploring tools that translate high level descriptions of biological functions into wetware</a> (e.g. DNA Parts) and also tools that create or connect hardware to those designs. Examples include automated design of microfluidic devices, connections to liquid handling robots, or connections to cloud automation infrastructure.</li></ol><p>There are no shortage of challenges. Problems that I thought would be solved by now are still very open ended. Customers are still using spreadsheets. Software tools are still one-off and <em>ad-hoc</em>. Lattice is excited to tackle these and other problems in 2019 and beyond. We are looking for talented, passionate people to <a href="https://latticeautomation.recruiterbox.com/">join our team</a>. The past twelve years have been incredibly exciting. I have no doubt the next twelve will be equally amazing.</p><p><strong>Douglas Densmore</strong></p><p><strong>President</strong></p><p><strong>doug@latticeautomation.com</strong></p><h3>Thanks</h3><p>There are many folks that I need to acknowledge along this journey. They include (in no particular order) Chris Voigt, Chris Anderson (and his lab circa 2007–2010), Adam Arkin, Jay Keasling, Leonard Katz, Kevin Costa, SynBERC PIs, Alberto Sangiovanni-Vincentelli, Nerayo Teclemariam, <a href="http://2008.igem.org/Team:UC_Berkeley_Tools">UC Berkeley 2008</a> and <a href="http://2009.igem.org/Team:Berkeley_Software">2009 iGEM teams</a>, Michael Kanaan, Swapnil Bhatia, Traci Haddock, Ernst Oberortner, <a href="http://cidarlab.org/team/">CIDAR Lab graduate students</a>, Kevin LeShane, Joshua Timmons, Bryan Der, Lloyd McCarthy, Ali Bagheri, Carter Wheatley, Erik Lindgren, Kadeer Amin, Raymond Chung, Samihah Islam, Alex Nero, Chris Batten, Nathan Hillson, Joanna Chen, Tim Ham, Jacob Beal, Chris Myers, Herbert Sauro, <a href="https://sail.bu.edu/">BU SAIL Team</a>, Barbara Densmore, entire Densmore family, Erika Parra, and Naomi Densmore Parra. Thanks Dad. Forgive me if I missed you! (please message me — doug@latticeautomation.com).</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=656e5f6812f3" width="1" height="1" alt=""><hr><p><a href="https://medium.com/lattice-automation/history-of-lattice-automation-656e5f6812f3">History of Lattice Automation</a> was originally published in <a href="https://medium.com/lattice-automation">lattice-automation</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>