How to accelerate Pharma R&D by turning researchers into data scientists

Jesse Paquette
Tag.bio
Published in
4 min readJun 27, 2018

--

Contrary to popular belief, data doesn’t create groundbreaking innovations by itself.

Pharmaceutical CEOs will laud their company’s data and computational prowess, but ultimately those assets are simply fuel for the intellectual engines at the core of biomedical R&D — human experts — biologists and biochemists.

Four years ago, we founded Tag.bio with the belief that:

  1. Innovation is a function of human capital. Systems which increase the capabilities and speed of human experts will have the biggest impact on the success and pace of translational research, drug discovery, and precision medicine.
  2. There is a significant human-scale bottleneck in data researcha last mile problem where researchers can’t codify and produce answers to their own questions without help from data scientists.
  3. We can dramatically increase the rate of discovery from data, and thereby drive innovation, by automating more of the process with software. That software must be easy to use, so biologists and biochemists can get immediate, useful answers at the very moment research inspiration occurs.

Imagine if — instead of doing a Google search, you had to send an email or schedule a meeting to ask another person to search the web for you. And then you had to wait days or even weeks to get back the list of search results…perhaps only to realize that your original question should have been asked in a slightly different way. This is how biomedical data research is operating right now. Accelerating that process is our mission.

Enabling faster discovery with software

Looking at the pharmaceutical industry, there are already a healthy number of software vendors and custom in-house solutions in the market — from the data-generic (Excel, Tableau, Qlik, SAS) to the data-specific (Spotfire, Genedata), to open-source programming languages and frameworks (R/Shiny, Python, Galaxy).

How is it, then, that our modest startup has been able to provide impactful, differentiating value in the Pharma world?

A modern user experience, simplifying and automating the process of data analysis.

Other software tools require significant training and/or encourage unreproducible manual operations. To put it another way — they make it too easy to do the wrong thing.

As a result of generic, hard-to-use software, biologists actually prefer to utilize bioinformatics service groups to ensure quality and accuracy of data analysis. However, bioinformaticians and biostatisticians are scarce, expensive, and can only work on one question at a time.

In contrast, our platform is explicitly customized for each dataset in collaboration with each research group. Simplified, bespoke workflows in Tag.bio give researchers a guided, step-by-step process to slice and analyze data within their comfort zone. Researchers get the exact features they need, without any of the other confusing, generic features intended for other data types and purposes.

This is a Tag.bio protocol — an analysis module designed for a breast cancer researcher to find biomarkers associated with patient mortality.

Agility, flexibility. Other software vendors and especially in-house systems have too much technical debt to adapt quickly (or at all) to new data and new questions. However, in the Pharma space, new data types and new research questions are a daily phenomenon. As data is augmented or new analysis methods are required, Tag.bio instances can quickly be adjusted with modular data connectors and protocols to match the emerging need. New analysis capabilities are thus available in our platform in hours, not months.

Insights-first, not visualization-first. Other software platforms tend to provide complex interactive visualizations and dashboards as their end result. This is a problem, as it creates confusing, inconsistent, subjective interpretations from data. After years of interviews with researchers, we focus instead on simple key metrics, precise visualizations, and clear, dynamic text relating each result to the context of the question asked.

Significant biomarkers and clinical factors are presented in a ranked, searchable interface. Arrows point the user towards follow-up questions.

Thus, the Tag.bio platform delivers atomic, objective, interpretable, actionable, reproducible insights (e.g. biomarkers, gene signatures, effective molecules, patient segments, drug targets) in the context of an uncomplicated and unmatched user experience for biologists.

Data is the essential asset for R&D-centric companies, but data is nothing without a useful interface for human researchers — the real idea machines powering Pharma. We aim to be that interface.

Please feel free to drop me a note if you have any thoughts, questions or feedback.

This story is published in The Startup, Medium’s largest entrepreneurship publication followed by 338,320+ people.

Subscribe to receive our top stories here.

--

--

Jesse Paquette
Tag.bio

Full-stack data scientist, computational biologist, and pick-up soccer junkie. Brussels and San Francisco. Opinions are mine alone.