Forum:How To Rescue The Life Sciences From Technological Torpor

Question

[William]

A call for more serious engineering in the life sciences by Bill Frezza, a semiconductor industry startup angel investor:

With the exception of DNA sequencing, which has enjoyed Moore’s Law-like improvements for a decade, drug discovery and development has failed miserably when it comes to harnessing a virtuous circle of ever increasing effectiveness and efficiency. In fact, it is moving in the opposite direction. This does not bode well for the future of a business under assault on so many fronts. Rearranging the deck chairs through waves of mergers and layoffs may temporarily fatten shareholder returns and executive bonuses, but such financial engineering will not cure cancer.

Read more at: http://www.bio-itworld.com/news/07/13/12/Skeptical-Outsider-rescue-life-sciences-technological-torpor.html

[User 59]

First off, don't repost copyrighted content from other sites:

For reprints and/or copyright permission, please contact Jay Mulhern, [phone], [e-mail].

Secondly this is not a question.

[Istvan Albert]

I reopened it as Forum - it is perhaps better to get the rants all in one place off our chest ;-)

[Istvan Albert]

Perhaps after removing the content leaving a short excerpt with a link it could be reclassified as Forum. After all we do have a similar post there.

Answer 1

[Damian Kao]

I agree with the article. Sciences have been converging for a while now. I don't think the various disciplines can just stick their heads in the sand anymore. Too many times have I heard undergrads or graduate students say, "oh, I can't do statistics or maths. I am just going to leave this to other people" or "learning programming is not worth my time". There are so many valuable perspectives you can gain on a problem by viewing it through other science's framework.

I am also a bit wary (as I think with many others) of this dependence on DNA sequencing as the bridge to other disciplines. DNA sequencing is just one dimension of a very complicated system. Getting high resolution of a single dimension is great, but at what point do the benefit gained from it plateau? If we are trying to model the motion of a ball based on it's x, y, z coordinates and we have a magic y coordinate machine that can give us extremely high accuracy to the 1000th decimal digit for our y coodinate. But our x and z coordinates are still whole number integers. How helpful is that really?

I am not claiming DNA sequencing is at the plateau phase already. I think ENCODE demonstrated that a variation of sequencing techniques (CHiP-seq, DNAse-seq, FAIRE-seq...etc) can produce some valuable data, but perhaps more can be gained by working on other high-throughput systems that will allow us to collect data on other aspects of the biological system.

I think there are plenty of other things in biological sciences that can be supplemented by works in other sciences.

• Image analysis: Cell counting, intensity quantification. Image-j and python image library has some cool fucnctions.
• Motion tracking. A paper in nature about burrowing behavior of field mouse (doi:10.1038/493284a) used this for some cool results.
• Datamining. So many papers are coming out these days, it's hard to keep up with it all. Being able to condense it all and get the information you want would be valuable. IBM Watson for academia?
• Data organization. Simple things like keeping lab animal stocks in order or reagents (although getting lab members to actually use it is a whole other problem).
• Databasing. Scientific results can also be databased with a controlled dictionary of terms (like GO). I work in the planarian regeneration field where various treatments (RNAi, chemical inhibitors) are associated with a particular phenotype during regeneration of the animal. A great database created by Michael Levin's lab (http://planform.daniel-lobo.com/download) recently surveyed available planarian literature and came up with a controlled system to categorized phenotypes and relate them to the experiment.

[Obi Griffith]

Well said! DNA sequencing is an incredibly powerful tool and there is still lots of juice to squeeze from that orange and further improvements and variations to be made. But, as you say it is just one (or several) dimensions in a massively (almost infinitely) complex problem. If I were a venture capitalist in biotech I would be looking to see what technologies are under development that will give us better real time imaging of biological systems and improved assays of epigenome, transcriptome, proteome, etc. Note, that sequencing can actually help with a lot of the latter but should not be our only approach.

Answer 2

[14134125465346445]

DNA sequencing has had ups and downs, but a much shorter history than drug discovery. DNA sequencing was pretty flat once capillary sequencing was established, and before 454 and Solexa entered the race. These and other companies only considered going from 1D (capillary) to 2D (slides/wells) when CMOS pixel-sensor technology went mainstream, something essentially borrowed from the digital camera and smartphone market. And now many of the new generation technologies are moving away from imaging slides and going straight to sensing electrical/charge differences directly in the wells or pores.

In relation to drug discovery and the patent cliff, do people have good statistics not on the number of approved drugs but on the impact drugs have on people? What I mean is that in the same way we can measure the DNA sequencing revolution by the amount of new DNA sequences produced and not the number of new technologies developed, can we measure drug discovery success by the deliverance to the world population instead of the number of approved drugs?

Answer 3

[Duff]

This is an interesting article to read in the light of the A Farewell To Bioinformatics forum posting bemoaning the state of bioinformatics. In that post we have someone trained in computing (I assume) venting his spleen (fair enough) about the imprecision in biology (esp molecular biology) and in the current post we have someone also trained in a highly technical and probably reproducible field complaining that biolgists are not taking advantage of technological and methodological advances in engineering to make their work more reproducible. C'est la vie.

However I think that one of the main problems with biomedical research is that it is not particularly reproducible. For example we sample (small sample usually) from a diseased population, carry out some assays, get back some noisy data and try to extrapolate from that. Setting aside the small sample problems if you and I both have asthma I doubt that we both have the 'same' asthma. The label 'asthma' is just that, a label for a constellation of signs, symptoms and processes that we often assume are driven by the same underlying cause. That's probably not always true (see here for example). Other polygenic diseases are also likely different at individual scales. The same review also makes the point that the pre-clinical animal models for diseases are mostly very poor models for the human disease. To my mind in drug development for human diseases we should be interested in the specifics of human disease not what we can generalise from pre-clinical models (indeed that does not seem to be working particularly well) and human biology is inherently noisy at the scale we currently operate at.

In light of this the application of more technology and, crucially, more individuals qualified to use that technology to biomedical and other biological problems should indeed be welcomed. We not only have to learn how to collect more human specific data but learn ways of stratifying within conditions (eg diseases) more effectively. The author of the article bemoans the lack of crossover from biologists willing to engage with technologists and I sympatise. However this works both ways. Those with more technological sense have to be willing to engage fully with the noisy nature of biological data and not necessarily expect the results of their techniques to be strongly reproducible across different samples. This noisey data seems to 'annoy' some of our more technically minded colleagues and there is likely no overarching model useful for all human biology.

The author ends on an optimistic note but the impact made by the researchers he mentions can only be measured once the compounds they are creating become available for treating diseases - early days yet. When I was a child we were promised flying cars and holidays on the moon!

In summary I think we are probably doing biomedical research 'wrong' to some degree and there does need to be change. I think that adopting techniques to study human disease in situ, decreased use of pre-clinical models and the application of technical advances to stratify individuals will have an impact. How that technical/biological dynamic develops will be interesting.