From CompBio
Revision as of 05:11, 22 April 2012 by Ram (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search



I consider all of us as students of science. Every student in our group has a great deal of freedom to pursue projects of interest to them and are not asked to do things they don't want to do (except in pursuit of the larger goal of doing science, i.e., rigourous controls are necessary to do good science, for example). Unlike most other professions, the goal in our academic programme is to foster independent problem solving abilities. I'm also a strongly antiauthoritarian and the beauty of science is to never believe in yourself or your ideas too much and always be highly self critical and questioning. As such, our group is run as a structured anarchy ("Structured chaos", "complexity", or "edge of chaos" are all terms that describe some of the science we do, much like life itself, and we also study how models like these are among the most efficient and effective). I however expect, as a condition of working together, that students recognise that "with great freedom comes great responsibility" (made famous by who?). One of those responsibilities is to leave the group significantly more learnt (learned) and wiser, surpassing everyone's expectations, including their own as well those of their mentor(s). The fact that students in our group have done so is one of the reasons for their success.

As a mentor, I believe in fostering an environment where each student's dreams and potential can be fulfilled to the maximum extent possible. While I am passionate about doing great science I also put the interests of the people who work with me above my own. I value the intellectual and personal relationships I have with my students extremely highly and I consider myself highly fortunate to be part of a group that not only does great science, but makes it a pleasure to do so.

More generally, I see the reason to do great science as not only personal but also as an existential responsibility. I believe that "science is the greatest achievement so far of the human race and its long term best hope for it's survival and enlightenment." As organisms on this planet who consume more than they produce, I see this as a way to restore the equilibrium.

On a more personal level, a close knit group is almost like a family. Not everyone has to agree. Disagreements are encouraged, but we need to end up working together in the end to achieve our common goals. If you felt a sense of excitement and passion when you first read this and continue to feel it, then keep at it! The rest will follow. If you think you that you could use a reboot, then by all means use all means at your disposal in the group to reboot.

Finally, even though I use the word "science", I don't see it as being distinct from philosophy or art or other categorisations of fields of learning. Constantly attempting to falsify one's discoveries is science. Our niche is that we do computational science, i.e., computational experiments that are well designed and rigourously controlled in a computational laboratory. (For us, a set of one or more computers is a system that we call a laboratory; our group therefore is referred to as just that, a "group", not a "lab" or a "laboratory".)

These ideas and others are expounded further below.

Doing great independent science is multifaceted

To do great independent science, it is sufficient and necessary to (listed in order of increasing difficulty) have great ideas, great solutions to (difficult) problems, followed by pretty great design and execution of properly controlled experiments. This has to be complemented by writing great, solid papers, and writing near perfect grant applications. Working smart and working hard is not an option.

Cynically, one may argue that what I am actually ranking isn't conceptual or technical difficulty but rather how enjoyable a given activity is. Furthermore, some of these activities have more to do with running a group than the actual science. This only partly true: writing great grant applications is more of a necessity as the world contracts and there is more competition for what seems like a smaller slice of the pie, and there aren't too many people I know who truly enjoy grant application writing. I however think grant applications are great tools for planning both the visionary and the down to earth details of one's scientific endeavours, from allocation of personnel and division of labour to learning by attending conferences that expose one to diverse ideas (though the cynical view may well be that such expenses are more about networking and salesmanship than learning). The changing nature of the academe and the commoditisation of learning means that funding for science is going to be nontrivial process for the next decades. The same can be said of publications, especially ones in high impact journals, which past a certain point have more to do with satisfying editors and reviewers than the quality of the underlying science. But proposal and publication peer review is still one of the most egalitarian, and yet rigourous, ways by which public and private resources are channeled for the common good.

I've always said that if you don't like the way the world works, you should dedicate your life to changing it. Science offers you the best hope of doing so on an individual level. Living in the ivory tower is something I am okay with but the results of what we do should be applicable to the real world. I advocate doing science in the most idealistic and pure manner possible. But being able to falsify the science done with real world applications improves the quality of the life for everyone lon this planet. The system isn't perfect and there is always room for improvement, something by its very nature that science, among all human activities, is most open toward. The social, political, and economic transformations occurring in the academe in general and science in particular, the massive amounts of data generated by powerful computing and high throughput technologies, means that multifaceted scientists engaging in collaborative interdisciplinary team efforts subject to review by capable peers free from corrupting influences is a necessity to do great science in the 21st century.

An idea is only as good as the number of the ways it can be falsified

The point is that, whenever we propose a solution to a problem we ought to try as hard as we can to overthrow our solution, rather than defend it. Few of us, unfortunately practice this percept; but other people, fortunately, will supply the criticism for us if we fail to supply it ourselves. --Karl Popper, The Logic of Scientific Discovery

As a scientist I wake up every day exciting about discovering ways I can falsify my own favourite hypotheses. The beauty of this is that the more you try to prove something wrong, even if it's something you love, and you come across a situation where your efforts to prove something wrong isn't working, then you can be somewhat reassured that you're on the right track.

Do not stop looking when you find what you are looking for

Related to the above: Most people stop looking when they find something, whereas we should keep looking because the next thing you look for might show everything you thought before was wrong (which is partly the reason people stop looking when they find something), and this happens most of the time, but the times it doesn't happen are the most exciting times.

Science needs to be done in a pure manner

Purity of the heart is to will one thing. --Søren Kierkegaard

Even though above I write about useful applications of science, basic science is is the foundation of modern mankind. Just like you can't build a great house without a solid foundation, doing applied science with an improper understanding or foundation of basic science I think is a recipe for disaster. Furthermore, not only is it the obvious---money, fame, politics---that can corrupt science, but a drive to do anything else other than solve the basic fundamental problems can also lead one to taking paths that lead to suboptimal or wholly inaccurate solutions. Science, as I write above, is one of the rare human endeavours where constantly questioning what is established is considered a good thing. As Karl Popper points out, people are easily enamoured of their ideas and are not prone to being critical of them, let alone be supportive and look hard for evidence in favour of opposing and contradictory ideas from others, especially when it originates from one's competitors who may not be likable for professional and personal reasons.

For this same reason however, even though it may be hard to be self critical of one's ideas, other people will gladly oblige and provide criticism when requested, as Popper observed. This concept is the reason peer review works. It also explains why a field such as protein folding was a largely dismal landscape littered with incorrect overtrained pet theories for several decades after the Nobel prize winning observation that demonstrated that at least for small well behaved examples, all the information needed for a protein to assume its tertiary (3D) atomic level structure was contained in its primary structure (or amino acid sequence), but rapidly started to demonstrate progress when a "competition" for the double blinded critical assessment of structure prediction (CASP) methods was instituted. Within three such biennial assessments over six years, significant progress in accuracy of prediction methods was observed in all categories of protein folding structure prediction algorithms, and within two decades, automated web servers were so sophisticated that leading experts declared the field to have matured, with new algorithms said to be capable only of incremental progress.

The protein folding structure prediction field had its wake up call with the first CASP1 where structure prediction algorithms did not do much better than what would be expected by chance. However, by CASP2, new algorithms that were honestly superior, as judged by independent assessors from the field, and the entire competition/experiment readily revealed what techniques worked, didn't work, and why. Seemingly promising and sometimes outlandish algorithms devised by well respected scientists with a strong publication history in the top journals did not necessarily perform well and some were among the very worst. Even if a person did not intend to "cheat", it was clear that repeated and iterative trial and error of software on limited test sets with known answers had led to a situation where information about the correct answers in the test sets had "leaked" into the algorithms with seemingly excellent results. However, when the algorithm was a given a novel protein that it had not been exposed to, then it failed to perform well.

The CASP experiments illustrate the human capacity for self deception. A good read on this phenomenon is Pathological Science by Irving Langmuir in 1953. Taken together, this represents that even seeming innocuous constraints can have a powerful effect (literally!) on the science being conducted. I personally try to eliminate self deception by not only submitting myself to peer review of all kinds (ergo the above philosophies) but also keeping a singularity of focus on the science being conducted and nothing else. While I would love to see a drug prediction made by our algorithms turn out to be a billion dollar blockbuster and/or save the lives of billions, I focus my will on only one thing and do my hardest to falsify our predictions. Money, fame, or even something as Nobel as saving billions do not inhibit me from being Popperian. Paradoxically, it is enlightened self interest to perform the most accurate, rigourous, honest, and best science. When it comes to getting at the scientific truth, there is no such thing as a failure unless one is being dishonest. Every attempt at falsification will either strengthen the original hypothesis if the evidence is corroborative, or lead to greater knowledge about the world if the evidence is repudiative.

Basic science needs to be verified in the most rigourous and useful manner

While I advocate doing basic science in the purest manner, one must also be ready to put on another hat and submit the final outcome to rigourous verification, meaning that it works and is useful in the field, in the real world, or whatever you want to call what's outside our Ivory Tower. People need to be able ultimately use what we do and go "wow". And what we do has to be more useful than an iPod or iPad or whatever toys we play with that are fun and help us on the path of enlightenment but are really in the end a small means and perhaps even a distraction. All this can be good and bad---technology is all about how people use it but it is not our right to coopt the consent of the people using it. It is their collective will that should determine what happens. Our pure goal should be to develop methods that save and improve the quality of the lives of billions of people (this is different from actually saving and improving the quality of the lives of billions people), not just seeing who can reproduce structure best or who can design a protein that does backflips. (This has value too but I've become cynical about just doing that and I feel our basic science should be pushed very hard and verified by a critical outside group, see Popper quote above, with a best, rigourous, and honest outcome.)

Do science that ultimately empowers people to improve their quality of life all by themselves

My long term dream (and this is a long, conceptually difficult, extremely exciting slog) is that people will be empowered to know their genome and metagenomes, to monitor them routinely, and to analyse and assess what to do to improve their quality of lives with the technology we provide. (It will of course be heavily based on computing.) So a person armed with their genome sequence information can predict what drugs will be most efficient for them, what kind of lifestyle to lead, what they should be not doing, all based on their genotype and mutations that are occurring in proteins that interact among themselves, with other DNA, RNA and small molecules. All this has to be done in the context of the metagenome which is driving a person's growth and genotypic and phenotypic evolution. A person can do this for themselves and their families. So it personalised individualised medicine at its finest, and all should be possible with a powerful computer in your garage that you can play and educate yourself with. A person's quality of life will then be in their own hands as much as possible, not in the hands of other entitles and complexes that human society constructs that form cliques to coopt the consent of the governed. (Yep, finally merged science and democratic political theory that led to the founding of the USA, inspired by the previous inhabitants who lived free here.)


This section is for core tenets that are brief and not (yet) described in great detail.

  • Pure science, art, and philosophy are indistinguishable from each other.
  • Science should be as simple, but not more simple, as is necessary; and should be as complex, but not more complex, as is sufficient. ("Science" can be replaced with "ideas", "solutions", "tasks", "programs", "publications", "presentations", and even "synthetic life".)
  • You are responsible for taking care of yourself.
  • You are born into this with certain resources that are the luck of the draw. One should give back more to the world than they take from it. Making the world a better place will ultimately make the world a better place for you.
Personal tools