The Vasco Da Gama Guide to Translational Research

In a knowledge economy the faster you know something the better.

Translational research and drug discovery are part of the knowledge economy.

How can you be certain that what you think you know is true? That is the billion dollar question.

With the ever expanding ability to do more and more complex analyses and generate more and more data pondering questions and seeking answers is no longer the exclusive realm of monks in a pea garden. It is a full blown collaborative effort.

 

Unimaginable Spaghettigram

 

In the head of every scientist, in every journal, in every library, in every lab in lab books there are multitude of hints. Hints about what might be the next big blockbuster drug. But.

They are only hints and nothing more. Why?

Complexity. While you might want it to be a simple network diagram, the reality is that there are many unseen influences on any potential biomarker or therapeutic target. You could make a reasoned guess.

This cytokine is triggered by that cell signal which is transcribed by that one transcription factor and they have all been shown to be present in elevated concentrations in the disease of interest. Has to be an important pathway.

Really?

Is it even reasonable to think that disease pathophysiology would be so simple. Of course if you read the intro to any manuscript or attend keynote talks everyone makes a sound case for ‘the’ target or ‘the’ biomarker. In reality given all the levels of integration possible and the millions of factors the pathophysiology of any disease must be an unimaginable spaghettigram.

Is it hopeless? Are the often talked about failures of the pharmaceutical industry due to a limit to what we can know?

Possibily.

If you talk about the ability of a single person to make these kind of judgements, then yes. The days of the solitary mad scientist are gone. However, even an auditorium of mad scientists brainstorming together won’t cut it either. Where does the hope lie?

With technology. The potential of technology, meaning computers, to help discriminate which hints to follow has not reached its limit. Far from it. There are however two ingredients that computers need.

Questions and data.

Even Watson, IBM’s supercomputer that beat Ken Jennings at Jeopardy, is useless without questions and data.

Humans are good at coming up with questions. Computers are good at analyzing data. Combine the two and you have something. It will be some time before we can ask a computer what is a biomarker for asthma and it will spit out an answer. We don’t have to wait.

We can even now exploit one of the greatest virtues of computers.

Speed.

Computers can churn through data quickly and put analysis at our fingertips. Yes, a proper study requires careful statistics, large sample sizes, and validation. Yet there are too many questions, too many hints and not enough time to do randomized controlled trials or careful laboratory experiments on all them.

 

Speed of exploratory analysis - case study

 

In the U-BIOPRED project transMART was tested using pilot data from a legacy study. One particular statistician was interested in getting an overview of the spread of values for a set of variables. To get the results of such an analysis it typically takes 2 weeks from the time it is requested until the data is ready. A statistician has to find the time, run the analysis, and generate the report.

transMART provided the answer in 2 minutes. That is 2100 times more efficient. An enormous gain in operational efficiency. Using a system like transMART makes exploratory analysis easy. Now a translational researcher can tick through all the questions he or she might have and learn where the effort should be focused. This is the value of putting translational research data into a knowledge management system. A huge value, but it won’t stop there.

 

Potential!

 

Kees Bochove, who is one of the core developer’s in the transMART Open Source development team via his role in the TraIT project, recently wrote a post about the current process of re-factoring the transMART core:

“it is fascinating to see how the disruptive power of the sharing and collaboration values behind the open source philosophy work their magic ways even in the IP fortresses of this world – the pharmaceutical companies.”

He goes on to point out that moving transMART to an open ecosystem that can support many different use cases is a challenge. Making transMART truly modular requires a lot of effort. The good news being that the work is progressing.

A new version of the clinical data layer has already been delivered. Bochove attributes the progression to the robust open source community that has developed around transMART.

eTRIKS is very much a part of that community and for good reason. eTRIKS aims to enable exploratory analysis of translation research data. This involves improving operational efficiency, assuring project legacy, developing standards, and enabling data integration.

What is most exciting is that with a modular transMART core there will be the potential for eTRIKS or any platform built off of transMART to be part of an analytical ecosystem that serves many different requirements.

 

Visionaries find routes to India

 

Christopher Columbus set out to the West. Vasco Da Gama to the East. Both were seeking a sailing route to India. Both were following hints. Da Gama was relying on reports of previous expeditions. Columbus on the concept that the world was round.

Technology makes exploratory analyses now more relevant than ever. It is a shift in thinking. A focus on a data in order to speed up discovery. A narrowing of the possible routes to a new therapy. Translational researchers need to do 4 things to achieve the same level of explorer greatness as Da Gama.

  1. Drive, adopt and adhere to standards: Data integration is much more efficient when collected under a common standard.
  2. Plan for knowledge management: Data management, knowledge management, IT infrastructure should no longer be considered the odd man out in research proposals ‘We’ll fund it if there is enough money.” It should be a first consideration.
  3. Avoid clinging to IP: If data can’t be shared across companies, institutions, and projects because of some theoretical concerns of legal departments, only a fraction of the potential of translational research will be reached.
  4. Ask questions and don’t be afraid to answer them with data: Many of the questions you have filed away for when you have the cohort, the money, or the will to do a proper study can already be answered by analyzing existing data.
  5. Engage in the growing translational research knowledge management community: Input from all stakeholders is needed and success is only possible with the achievement of a critical mass.

The time of data driven exploratory analyses is nigh. Are they part of your strategy?

  • Kees van Bochove

    Hello Scott, Great post, very eloquently written! I couldn’t agree more about your first point. If translational scientists would put just a little more effort into standardization, this would have a huge pay-off in the long run. Of course having an IT infrastructure which facilitates those standards and which also demonstrates its value and usefulness, such as tranSMART, is an important step to drive the standardization effort. But equally important is the mindset change from what you call ‘clinging to IP’ to sharing not only results but also data.

  • http://horizon2020consulting.com/ Scott Wagers

    Kees thanks for the comment. Your point about having a platform to demonstrate the value of having standards is a great way of looking at it. Indeed, from my experience when you move past people who work with data or data management on a regular basis, the value of having standards becomes rather abstract. Give them the keys to a transMART implementation with a few good datasets and let them find out how quickly they can answer questions they’ll be impressed. But it is important that it’s clear that its not magic, but rather the payoff for proactive work of setting standards, collecting good data, and curating it.