How bioinformatics can transform cell-based therapeutics


Therapeutic products have changed over the years. The discovery of penicillin  in 1928 marked the dawn of the ‘small molecule’ era, in which scientists began to design and develop chemicals with known properties, chemicals that can alter human health. The ‘antibody’ era kicked off in 1986, with the approval of the first monoclonal antibody, muromonab-CD3, for the rejection of kidney transplants.

The most recent class of therapies – cell-based therapies – came to the forefront with the demonstration that cells that have been manipulated outside the body can have a long-lasting therapeutic effect in people with leukaemia. The first patient treated with that first cell therapy, now marketed as tisagenlecleucel (Kymriah), has now been cancer- free for over a decade.

Although they were originally developed for certain types of cancer, cell therapies are moving into the mainstream. There are now nine cell therapies approved and hundreds more in clinical trials: cell-based therapies can now cure many leukemias and treat some rare diseases. However, major hurdles still need to be overcome for cell therapies to become pillars of medicine in multiple therapeutic areas.

 Many existing cell therapies use autologous sources (meaning that the patient’s own cells are manipulated ex vivo and reintroduced, so you can’t make a therapy that could be used to treat lots of people).

In 2006, seminal work from Shinya Yamanaka and colleagues showed that fibroblasts, cells commonly found on our skin, could be reprogrammed to resemble stem cells with unlimited proliferation ability, and that these cells can generate almost any type of cell. These induced pluripotent stem cells (iPSCs) from one patient could be used to treat another, and have paved the way for the development of multiple novel therapeutic alternatives, addressing a diverse range of diseases. However, despite the immense potential, the lengthy protocols and expensive reagents used to make these cells (and the sub-optimal products that result) mean that relatively few clinical trials evaluating iPSC-based therapies have been initiated, and none has become a therapy that’s available to patients, yet.


A key prerequisite for using iPSCs as therapies is finding ways to efficiently derive pure populations of specialised cells (e.g., brain or heart cells) that could be used to treat different diseases. Over the past two or so decades, numerous protocols have been established. In these, scientists add molecules to iPSCs grown in the lab to turn those cells into other specialised  cell types that might be useful as therapies. They’re taking what we know about embryonic development - that cells need to be exposed to specific molecules at specific times - and trying to mimic those processes in the lab. So far, scientists have successfully made cells that could be used to treat diabetes, certain types of blindness, and Parkinson’s disease. However, it takes months to make cells this way, it’s expensive and complicated, and doesn’t always produce cells that could be used as therapies.

Forward programming could be a  solution to overcome the limitations of classical differentiation protocols. This method is similar to the approach Dr Yamanaka used, in which he  expressed four transcriptional factors (now known as the “Yamanaka factors”). Unlike classical differentiation, which requires understanding the complex biological processes during human embryonic development, forward programing relies on transcriptional profiling of the starting and target cell types to identify the right combination of transcription factors that can convert cells. This approach has been used to differentiate iPSCs to neurons with nearly 100% efficiency through forced expression of a single transcription factor.

So far, it has been possible to generate only a limited range of cell types with high degree of purity using forward programming, which highlights the importance of identifying effective transcription factors for converting cell fate.  Thanks to recent advances in ‘omics, including single-cell transcriptomics, we are now able to generate, assemble and integrate large datasets to computationally build a comprehensive single-cell reference atlas. If we can build a model to understand this atlas, we could uncover the relationship between transcription factors and gene regulation.

The opportunity here is enormous, and is one that we are actively working on at Deep Science Ventures. Sina Atashpaz joined us 3 months ago, with the broad remit of identifying innovative solutions to overcoming the challenges with iPSC-based cell therapies. By evaluating the constraints within the field, he has identified that, by developing analytical systems to evaluate pluripotent stem cells and their differentiated derivatives, we could use that information to optimise differentiation methods. This would enable us to generate numerous therapeutic products at scale and revolutionise our ability to develop cell therapies across a variety of diseases with high unmet need.

Sina is now hiring a co-founder to found a company, to move cell therapies into the mainstream by finding more efficient ways to generate cellular therapies. If you are an entrepreneurial bioinformatician or computational biologist, who is interested in founding a company within this area, we want to hear from you! Please see more details on the role and the application process on this job description.