Getting Real: Genetics research, genomics and gene-tech

17th December 2020 about a 9 minute read

Read the latest installment in our short story series by Stephen Palmer here. 

The title of this month’s story, Genomancer by Stephen Palmer, is a nod to Neuromancer by novelist William Gibson. By switching neuro- to geno- the author of our sci-fi story highlights the focus on genetics. Given that Alzheimer’s is a main theme of the short story, neuro would also have been appropriate.

The explores how genetic sequencing and artificial intelligence (AI) could be used to predict diseases, in this instance Alzheimer’s dementia. Genetics alone could form an entire blog post, as could Alzheimer’s or indeed diagnostic applications of AI. However, I have chosen to focus on the history of genetics research, methods for looking at genes and the implications of consumer focused “gene-tech”.

History of genetic research

Genomancer touches on themes which are being increasingly explored in academia, industry and policy environments. As we will see later, genomics is a major part of the UK Government’s R&D strategy. The last 70 years has seen enormous steps forward in understanding of genetics. Structural discoveries in the 1950s about genetic material by Rosalind Franklin, James Watson and Francis Crick laid the foundation for the improvements in techniques to investigate the function of genesand, more recently, to manipulate and alter genes and their associated function. (Gene manipulation would certainly warrant a post of its own, although we’ve touched on CRISPR and designer babies previously.)

In the early days of genetic experiments, species of plants and animals were bred to look at physical characteristics, for example in Gregor Mendel’s experiments with peas in the 1800s. Charles Darwin observed similar things in finches and over time researchers applied his theory of natural selection and began to identify adaptation of functions, with a biological vehicle to transfer characteristics between generations. While the concept of genetic transfer had been described through the 19th century, specific details about chromosomes and genes came later, in the 20th century. Chromosome were first described in 1888, but the demonstration of genes being located on chromosomes came in 1910 and the term “Gene” was only coined by Wilhelm Johanssen in Denmark the year before.

If we think about this months’ story and a link between diseases and genetics, it is interesting to consider that cancer was first associated with heritability and chromosomes in 1902. (Figure 1. In this article is a timeline of some of the significant milestones in genetics research.)

Being able to visualise genes and genetic information was an enormous step forward, for academic research, as well as for the development of modern biotech companies. Imaging the structure with x-rays paved the way for the discovery of the deoxyribonucleic acid (DNA) alpha helix and subsequently the structure of individual base units. From the 50s onwards, different methods for gene sequencing in order to figure out the order of the base units in DNA have helped researchers understand how individual genes can code different physical features or functions, as well as the links with diseases. Genes for a code can be translated to make units of protein. The structure of these proteins makes up our physical appearance and function, we will come back to this when we talk about AI advancements.

The development of the polymerase chain reaction (PCR) is used to multiply genetic material to help detect genes and make it easier, and faster to determine which genes are present in a sample for testing. Initially this was done by hand in temperature control baths, but now machines automatically cycle between temperatures for the different steps which amplify the material. The PCR has been regularly featured in the news throughout 2020 as the PCR is part of the approach for processing COVID-19 in swab samples.

Genes change over time through mutation. This is normal and it’s what causes differences in appearance or function. This is how natural selection works, but it’s also how genetic diseases occur. There are many different genes linked to diseases, for example these can be risk genes which we know many of for different types of cancer, or they can be dominant genes which cause a disease like Huntington’s. From the perspective of the story this is very important as there is a big difference between a risk related gene and a gene which causes a disease. When it comes to Alzheimer’s, there are many genes associated with a greater risk, although unlike Huntington’s disease, no gene associated with Alzheimer’s has a yes/no binary cause.

In recent years, the cost of genetic sequencing has reduced considerably. This is partly because The Human Genome Project has made genetics research more accessible, coordinating efforts to fully sequence the full complement of human genes for the first time. This has enabled genetic sequencing enterprises to develop and there are now several services such as 23andME, which will sequence your full genome. These services can provide information about your geographical genetic heritage and certain medical details. However, there are certain considerations to reflect on. Once the information is known, you are not able to go back, which has caused upsets about paternity, as well as knowledge about increased risk of diseases causing worry. It is also worth considering how the data may be used in the future. If we consider the story, a company selling consumer sequencing services may pivot to a new domain, concerns have been raised in the past about how genetic sequencing could be used by medical insurance companies to tailor costs to risk. This however is a huge topic which would warrant a dedicated post.

A major area of research into genetics is the translation from a genetic code to a 3D formed protein. There are lots of interactions between proteins, which when translated from the genetic code make up a “beaded necklace” structure. Each protein, or bead, can interact with the others to create a complex folded shape. It’s very difficult to predict what this folded shape will be, but in the last month, DeepMind has announced that it’s AlphaFold2 AI project has dramatically improved the success of predicting this 3D structure from the basic genetic code, by 30%. This is a significant breakthrough and is really promising in terms of offering a better understanding how a gene links to protein which then makes up physical characteristics and functions.

Detection vs diagnosis

Beyond genes and genetic material, the difference between identification or detection of a disease and clinical diagnosis is subtle, but important. In a research domain, if a scientist is looking at features of a disease, symptoms, or a biomarker for an illness or condition, it is important to make a clear distinction between something that can identify the disease in a comparison with healthy controls and something that a physician would use as a clinical test for diagnosis. It may seem pedantic to the extreme, but as with the differences between genetic risk for Alzheimer’s and a gene determining whether you will or won’t develop another neurodegenerative disease such as Huntington’s, features or symptoms for detection is not a formal diagnosis.

A good example here are the experiments looking at the early detection of Alzheimer’s disease using keyboard typing is an interesting research study, however it is not currently a diagnostic test or validated method for diagnosing Alzheimer’s. You could perform in a certain way in the experiment and analysis of your data may categorise you in a certain way, but without the full evaluation with a neurologist and diagnostic testing, it would not be able to tell you anything concrete. This is the case for many tools and approaches analysing large data sets, on a whole host of different clinical conditions as well as general behaviour, or interaction with a smart phone or piece of wearable technology. Several AI drug discovery companies are developing approaches to make it faster and cheaper to identify drug targets and take said drugs to market. Benevolent AI have taken big steps forward in using AI and genetic information for drug discovery, most notable with amyotrophic lateral sclerosis (ALS), using funding from the “ice bucket challenge” to fund the development of a novel drug and take it to market, in this instance $115M of funding was provided to identify a suitable drug candidate and get it to the stage of clinical trial. Pharmaceutical research and development is extremely expensive and estimates of cost have been made at over $2Bn per new drug taken to market.


The Government has been developing its genomics strategy and this year announced its national genomics board. The “Genome UK: The future of healthcare” strategy, published in September demonstrates a clear  commitment to the importance of genetic research. It highlights the importance of industry collaboration and development of a “vibrant start-up economy”. Currently, there are 104 UK companies listed on Crunchbase as active in working on genetics or on gene related products and services. Some of these are involved in AI and data analysis, however the majority appear to be providing either sequencing services or reagents.

With this focus on genomics, big data, and AI as main areas of focus for UK research and development in the coming years, we will likely see a proliferation in this sector niche, potentially with many more consumer facing products entering the market, potentially some with a similar implementation to the Cassandra application in the story.

As we have already discussed, new analytical approaches and methods for detection of genetic diseases are helpful in research and clinical science, enabling scientists to know more about diseases so they are better equipped to develop treatments and cures. However, developing such approaches as a consumer facing application should be very carefully considered. In Genomancer we see a very limited timeframe of Alice and Holly in this regard and we don’t know exactly what data the app has access to. However, it seems irresponsible and unscrupulous to be deploying a product to diagnose Alzheimer’s based purely on genetic data, without involving a a trained clinician in carrying out the diagnosis and working through the implications of a diagnosis with a patient and their family.

In reality, rather than a notification pinging from an app, diagnoses are made by a clinician. Similarly, in the case of genetic disorders, genomic counsellors help with the emotional burden of such news.