Under the FDA’s definition, GMO crops have been around for over 20 years (though banned in the UK and many other countries in the EU), but GMO humans have remained a sci-fi fantasy up until this year, with the shocking news of a scientist in China claiming to have delivered twin babies with CRISPR edited genes rendering them immune to HIV, opening a new chapter in the world of human gene editing.

Gene Editing with CRISPR

Gene editing is the process of modifying DNA within a cell, making cuts at specific sequences using enzymes called ‘engineered nucleases.’ These enzymes are made up of two components: a nuclease, responsible for making the cuts, and a DNA targeting part (this is usually RNA based or protein based). Once the cut has been made, the cell is able to recognise and repair the DNA, joining the severed parts back together. The first nuclease technology, known as Zinc Finger came about in 1991 and the protein based DNA slicers themselves have been known of since the 60s. The issue with this method is the lack of specificity when making cuts, which can have very undesirable consequences.

Enter CRISPR. Discovered ‘accidentally’ by Jennifer Doudna and Emmanuelle Carptentier in 2009 whilst studying bacterial defence mechanisms, CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats) is a naturally occurring defense mechanism in microbes against viruses. Using a nuclease called CAS-9 along with small guide RNAs (sgRNA), they are able to enter the genome of a virus and deactivate their genes with great precision. Unlike other methods of gene editing, the same CAS-9 nuclease can be used for all targets, whilst different sgRNAs are used depending on the target. As well as its stellar precision, what also separates CRISPR from other methods is its speed, affordability and ease of use, which makes it more feasible for clinical use. This is why there is so much hype around CRISPR and why it could potentially be used on a much bigger scale.

Until very recently, gene editing has been limited to use in scientific research in non human species, or non embryonic human cells. But it seems as though the bar is being pushed — just a few months ago, a team announced that they successfully wiped out the African mosquito (in a small controlled environment) using CRISPR to render them infertile, generating a gene drive that eventually wiped them out. If this is actually performed in the real world, millions of lives could potentially be saved, but many question whether such an act could have many unwanted consequences.

Indeed, the discussion of gene editing is constantly followed by questions of ethics. As mentioned previously, it was recently announced that researcher He Jiankui delivered twin babies with CRISPR edited DNA in China. This caused an uproar in the scientific community, with even the CRISPR co-inventor herself describing the move as ‘disturbing and inappropriate.’ Albeit controversially, China has always taken the lead in genomic research. It is the only country where testing on embryonic cells is legal, but taking the embryo to birth takes things to another scale that the majority of the scientific community is in total disagreement with due to the lack of knowledge on the risks of associated with modifying human DNA, especially in otherwise healthy babies. It was described by Joyce Harper at Women’s Health, UCL as ‘premature, dangerous and irresponsible.’

Genome Sequencing

Regardless of the ethical debate, more and more people are interested in the inner-workings of their DNA, which brings up the question of whether we should be free to do what we want with our DNA and that of our offspring, despite the potential consequences.

Since the human genome was sequenced in its entirety for the first time in 2003, the cost and ease of doing so has dropped tremendously. The initial attempt cost around £2 billion. By 2008, the cost had dropped to under £1 million and now, you can have your entire genome sequenced for under £1000. In the last 5 years, many websites such as 23andMe have popped up, offering users the opportunity to find out more about what lies within their genes for under £100, by simply sending off a spit sample. It is worth mentioning that these sites do not actually sequence your entire genome as this would be far too expensive and time consuming for a casual user. Instead, they use a method called genotyping — the process of determining which genetic variants one possesses. It picks info at specific DNA sequences rather than the entire genome so it’s not as complete as sequencing your entire genome, but on the plus side the variants are well studied and you can learn some interesting things about yourself.

Despite the limitations of genotyping as opposed to genome sequencing, a lot can be done with it. There are many DNA analysis apps available, for example Athletigen, which takes your sequence from 23andMe and provides you with information relating to health and fitness, like the best type of diet and exercise for you.

If you’re interested in creating your own app, there are some useful APIs like this very cool one from genomelink.io, to facilitate this. As shown in their frequent hackathons, there are many ways you can use make use of this API to create some interesting things.

There is definitely a growing amount of people increasingly interested in not just knowing about their DNA, but modifying it. This can be observed in the very active ‘biohacker’ community. On the less extreme end, some of these people are immensely dedicated to combating their genetic makeup through extreme lifestyle changes and supplementation. On the other end, there are guys in their basements injecting CRISPR plasmids into their abdomens with the hope of delivering it into their cells and changing their DNA. As absurd as this may sound to some, this all illustrates a shift in people becoming increasingly curious about the inner workings of their bodies, taking ownership of it and pushing nature’s limits to suit their desires.

Like code is quickly becoming a language more people are familiar with, perhaps the same is also happening with DNA. And just as the increasing ease of learning to code creates the opportunity for malicious use, knowing how to interpret and modify DNA could do the same. Just as He Jianku went against ethical norms by delivering those twins, there could be many more ‘rogue scientists’ hoping to do similar things — either for themselves or for financial gain. An article in Stat discussed the possibility of scientists opening illegal IVF clinics to offer CRISPR editing. Paul Knoepfler, a stem cell biologist at UC Davis argued that there’s a very real possibility of this, given that similar things are already happening with many clinics offering non FDA approved services, whereas Hank Greely, a Stanford law professor believes that it’s totally unfeasible. Regardless of the debate at hand, human gene editing is fast becoming a significant part of our reality and it does bring up a lot of questions as to whether the laws and regulations in currently in place can handle this.

I would highly suggest checking out the Gene Literacy Project, it’s a great site that covers all aspects of genetics. Just as there is a great push for everyone to learn to code, there is also an increasing need for gene literacy.

Mind Reading with Brain Computer Interfaces

Since my first post last month about General Adversarial Networks, my interest in neural networks has persisted. In that post I discussed how AI can be used to generate original images. In this post, I’ll be exploring another fascinating way in which neural networks can be used — to extract information and facilitate communication between brains and computers, and other brains.

You may have noticed that ‘mind reading AI’ has been a hot topic in the media this year, from news on Facebook exploring technology to write posts with your thoughts, to Elon Musk recently announcing his latest venture with Neuralink, a ‘neural lace’ to be implanted under the scalp to enable one to interact with machines.

Despite the recent news surrounding it lately, ‘mind-reading’ technology has actually been around for decades. There are two commonly used methods for the purpose of reading brain patterns — electroencephalography (EEG) and functional magnetic resonance imaging (fMRI). However, until the last couple of decades, it’s been quite tricky to make use of the data produced by these methods. Thanks to the advancement of artificial intelligence, it’s becoming far easier to decipher the data. The combination of these methods forms what is known as a brain-computer interface (BCI).

fMRI is the more recent invention of the two. It indirectly tracks neuronal activity by tracking blood flow throughout the brain. When neuronal activity in a particular area of the brain increases, there is an increased demand for oxygen in that area and as a result, blood flow to the area increases to deliver the oxygen. This increased blood flow is picked up by the fMRI. This shows us what area of the brain was working at a particular moment, which can provide some clues as to what is going on in the brain, however it’s not easy to figure out what exactly the person was thinking at that particular moment.

Enter neural networks. As discussed in my previous post, neural networks are systems designed to mimic human neuronal activity and progressively improve at a task through algorithms and training sets. Neural networks can be used to decipher patterns in fMRI scans and associate active areas of the brain activity with particular thoughts or activities. In order to do this, it needs to be trained with tens or hundreds of thousands of fMRI scans, labeled with what activity was occurring at the time. For example, feeding it with several images of a fMRI scans whilst looking at a particular image. Eventually, it will be able to recognize by itself the brain activity taking place whilst looking at that image.

There have been some great studies demonstrating the use of neural networks and fMRIs. Zhongming Liu and his team at Purdue University trained a neural network to choose from a selection of 15 options (including a face, bird or plane) of what a person was looking at. In order for it to be able to do this, they trained the network with several fMRI scans of human brain activity whilst looking at each of those things. The success rate was about 50%, which is fairly impressive.

fMRI based BCIs have also been used to recreate videos based on what a person was watching at the time of the brain scan. Jack Gallant and his team at UC Berkeley set this up by recording brain activity while a subject watched hours worth of movie trailers. The brain activity was mapped with the content of each video frame that it corresponded with, which was then used to train the neural network. Through this, the neural network learned to predict the image someone was watching based on the recorded brain activity. Below is a video showing the most accurate results it produced. If you’ve read my previous post on generative adversarial networks, you will find that this sounds quite similar. It is in fact a different type of generative model in this case, but generative adversarial networks have indeed been used in other cases such as this study, where the system was used to generate images and videos based on fMRI input.

Similar work was also performed by Yukiyasu Kaminani, published in an article earlier this year, using a neural network to reconstruct images of letters a person was looking at, based on fMRI activity (image below). Kaminani has also worked on identifying objects in dreams using the same technique.

Despite the amazing work that has been produced using fMRI, there are limitations. It is incredibly expensive. The method itself (lying still in a large machine) is totally impractical for everyday use. It’s also quite difficult to obtain the amount of data required to train the neural networks.

As an alternative, an electroencephalogram (EEG) can be used to read brain patterns instead. Whereas fMRI records images of brain activity based on blood flow in particular areas of the brain, EEGs record the actual brain signals. A cap (similar to a swimming cap, with chin straps) covered in conductive electrodes is placed on the scalp, which record the electrical currents that occur when neurons communicate.

Tracking these occurrences appears as waves on a graph. There are 5 different brainwaves typically used in EEGs (though there are many, many more), and they are each associated with different types of brain activity.

Electroencephalography has been around for almost a century, but combining its use with a neural network has massively expanded its application. There are a couple of drawbacks with EEG that can make it difficult to read patterns. Firstly, as bone is a poor conductor of electricity, the skull interferes with the reading. It’s also quite difficult to track specific brain activity due to unwanted extra noise from other activities. One way to get around this is by getting the electrodes closer to the brain by surgically implanting it under the skull, which is how EEGs were originally designed by its creator Hans Berger, and is what Elon Musk is working on with his new venture, Neuralink, in the form of neural lace — an ultra thin mesh, implanted directly onto the brain. As bizarre as this sounds, it has already been performed successfully on mice. Of course, this most likely won’t be an option in the near future for most consumers, but the less invasive cap option, coupled with a good neural network can extract the useful information and is far more feasible for consumer use and has produced some amazing applications. EEG based BCIs have been used to control wheelchairs, neuroprosthetics, drones and even humanoid robots.

More recently, BCIs are increasingly being researched for use in gaming and social media. Last year, Mark Zuckerberg wrote in a Facebook post:

‘We’re working on a system that will let you type straight from your brain about 5x faster than you can type on your phone today. Eventually, we want to turn it into a wearable technology that can be manufactured at scale. Even a simple yes/no ‘brain click’ would help make things like augmented reality feel much more natural.’

As if that wasn’t bizarre enough, BCIs are also being used to facilitate communication between human brains. Rao and his team at University of Washington created BrainNet, the first ever multi person non-direct brain to brain communication interface. This interface allowed three people to communicate via BCI to play a game of Tetris. Two of them were assigned the role of ‘senders’, delivering instructions to a third person, a receiver, of whether or not he should rotate a Tetris block.

The only way the receiver could play the game successfully as by receiving information from he senders, as they could see the Tetris pile at the bottom, and the receiver could not.

The two senders wear EEG caps to transmit their signals. The brain signal is created by them looking at a flashing LED that says yes (do rotate), or another one that says no (i.e, do not rotate). Looking at either of these signs produce a signal, which is sent to and recognised by the neural network via the internet.

The neural network converts the raw EEG data to a single command, which is then sent over to the visual cortex of the receiver using transcranial magnetic stimulation (TMS21). If this works correctly, the sender’s ‘yes’ would appear as a phosphene in the user’s vision (the strange patterns that appear when you rub your eyes) to the receiver, and a ‘no’ would not produce sufficient energy to trigger the phosphene. The receiver is also equipped with an EEG to perform the task of rotating (or not rotating) the Tetris block by looking at an LED yes or no similar to the senders’, which is then taken up again by the neural network and actioned on. They managed to complete the game with about 70% accuracy, which is quite amazing.

As you can see, reading thoughts is no longer a science fiction fantasy. It is quickly becoming a reality. Though the technology is still in relatively early stages, it can only progress further and the possibilities are really exciting to think about. Being able to store your own thoughts on a hard drive and put them back in when needed. Revisiting your dreams and sharing them with others. Perhaps we could even communicate with animals one day! Of course, as with most technology, there are a lot of potential negative implications which I’m sure you can think of, but regardless, I’m quite excited to see where this goes. It’s also pretty cool to think of how we’ve essentially created a system to the mimic human brain, and then use it to decipher the human brain.

Brain Computer Interfaces was originally published in DataDrivenInvestor on Medium, where people are continuing the conversation by highlighting and responding to this story.

For my pairing project this week, we decided to create a CLI application for users to look up local artists and exhibitions. There are so many platforms around to find out about local gigs and club nights (Songkick, Dice, Resident Advisor) but besides the awesome website Artsy, I haven’t come across many other widely used local event apps in the art scene. From experience, I can say most art newbies will probably have a bit of difficulty finding out about local events if they don’t already have friends in the know, so I figured building a little program to facilitate this would be pretty cool.

Working on this project piqued my curiosity about other ways in which the art and tech worlds intertwine and I recently came across an article about French startup Obvious auctioning off bespoke paintings created by AI, for thousands of pounds. Though there has been a bit of controversy surrounding this, which you can read a bit about in the article, what really intrigued me was the technology behind the images — Generative Adversarial Networks.

Generative-huh? What’s that?

In short, generative adversarial networks (GANs) fall under the umbrella of generative models — a form of deep learning used with the aim of creating unique data (most commonly, images) with a likeness to the training set it was provided with. GANs were created in 2014 by Ian Goodfellow, now a lead research scientist at Google. Over a pint of beer one night, he came up with the idea as a solution to the lossy and blurry images being produced by the commonly used generative models at the time. What separates GANs from the others is that they are comprised of two neural networks — a generator and discriminator, working in an adversarial, yet symbiotic relationship to achieve the task of producing new data. It’s a very exciting concept that has applied in lots of interesting ways and is often touted as the future of deep learning.

Neural networks — A set of algorithms made to mimic the behaviour of neurons in the human brain.

How GANs work

To relate to the topic of art, the relationship between the generator and discriminator is like an ongoing battle between an art forger and an art dealer. The inexperienced forger presents a terrible Van Gogh replica to an art dealer is who is pretty new to the game himself, but even he could easily recognise the poor attempt at forgery.

Undefeated, the forger tries again and again to produce a replica to fool the art dealer. Each time the fake is spotted, the forger works hard to improve until eventually the art dealer can no longer distinguish the real from the fake.

In ever so slightly more technical terms, the generator is provided with a training set of data, all belonging to a particular class (for example, a set of Van Gogh paintings) and attempts to produce a similar image with random noise generated from the data set (this is the case with Vanilla GANs, other types of GANs, such as conditional GANs are given a less abstract input). This image, along with the training set, is presented to the discriminator, and the discriminator returns a probability of 1 (yes) or 0 (no) that the image belongs to the class of the training set. The generator takes this answer and makes use of it to learn the distribution of the images in training set and attempt another go at creating a convincing image. With each response from the discriminator, the generator gets better and better at learning the distribution.

Lets simplify this a bit

Faces are a good example. In supervised training, a neural network is presented with images of thousands of different faces and eventually it starts to recognize the common features with each face — eyes, a nose, a mouth, until it’s able to recognise a face on its own. With a GAN, the process is similar, except the generator is assisted with nudges in the right direction from the discriminator. It’s like a game of hot and cold! One player is blindfolded with the assistance of 1 of only 2 possible answers from the other player to guide them in the right direction. When you think of it like this, you could say that the the discriminator is actually encouraging the generator to reach its goal rather than fiercely competing against it, like one may immediately assume. It’s quite sweet, actually.

GANs in action

Below are some awesome examples of how GANs have been used to generate images and videos in various forms.