Temple Smith is best known in the bioinformatics community as the co-developer of the ubiquitous sequence alignment algorithm that bears his and Michael Waterman’s names. The algorithm, published in 1981, is widely considered the standard underlying tool for most DNA and protein sequence comparisons. Smith, who earned a PhD in nuclear physics before turning to biology, also co-founded GenBank at Los Alamos National Laboratory, where he helped develop a statistical mechanical model of the Lac operon system.
Now at Boston University’s BioMolecular Engineering Research Center, where he has served as a professor and director since 1991, Smith is a frequent speaker at computational biology conferences. He is also the co-founder of Modular Genetics, a gene and protein engineering company based in Cambridge, Mass.
Most recently, he was presented with the International Society of Computational Biology’s senior scientist award at the annual Intelligent Systems for Molecular Biology conference, held jointly with the European Conference on Computational Biology in Vienna last month [BioInform 07-27-07].
Before giving his keynote address and receiving his award at ISMB, Smith sat down with BioInform to discuss what lies ahead for computational biology, why he doesn’t like the term “bioinformatics,” and what the new guard can learn from the old.
Will bioinformatics be a thing of the past in 10 years?
I do not know what the term means! Someone pointed out … that it was first put in quotes by Jean Michel [Claverie], [when he said] bioinformatique …As far as we know that was the first time it was used, the first Waterville Valley meeting. I feel that organizing those three meetings was my major contribution to the field.
So, what is next for computational biology?
I will say that there will be, over the next few years, considerable consolidation [in the field] … It might be five years, might be 10. We don’t need 30 different copies of each different algorithm. And that [consolidation] will be partly driven by both education and research needs; that is, to some extent, [that] everybody uses Blast, for example, and you will find more of this kind of consolidation — standardized software packages that are tools of every biologist. And the part of this field that is useful will split between applied biology and, basically, the medical sciences, and my guess is we will no longer be a pseudo-discipline.
How will the split affect the field?
It’s not a field. [J.B.S] Haldane was interested in a lot of [the same] problems we are, [such as] analyzing genes, [and in] 1919 … published over 200 papers in what today you would call ‘mathematical genetics,’ [or] ‘mathematical medicine;’….D’arcy Thompson … wrote those wonderful books [such as On Growth and Form]…Sewall Wright was writing his books in his 80s in Wisconsin … [These books are] still complicated enough that half the people in the world could not work through them [today], including yours truly.
What’s changed is the data, not the problem … The first [computational biologists] were on-hand calculators.
What I mean is, if you were in a biology lab or a molecular biology lab back in the ‘60s or ‘70s you blew your own Blast. Did anybody even know how to do that? I can still do that. Over the last 10 years they’ve talked about computational biology sort of being a service discipline to biology … [and the computational biologists] resent that, they want to invent and so on.
Computational biology only works if you work as a team. For example, in modern geology you have chemistry, physics, evolutionary biology, … mathematics, and lots of computer modeling. There is a true interdisciplinary thing. Does anybody talk about geoinformatics? No! They do not… The problems are biology. So you have the computer and mathematical tools. You teach courses in biology departments on how to use tools that are already standard, like Blast.
Usually, an introduction to crystallography [is included] in biology departments. Lots of biology is medically oriented, [and is] somewhat different from the history of life and so on. You already see that split. [A] person who gave [a] talk on mathematics [at the ISMB conference] did not have most of the molecular people in the room. You already see that … and I think that will become more common because the data and the computer tools sort of get specialized.
What challenges lie ahead?
Most people would argue that one of the challenges is trying to figure out ways of doing modeling that allow you to actually model the complexity of living systems at the molecular level.
Who is ahead?
For molecular simulation it’s a group at Los Alamos. And one key person modeling regulatory systems — [the] most detailed [ones] — is Jim Collins at my university … realistic simulations so you can do in silico mutations and changes and maybe simulate some evolutionary events. But … my main message today is …lots of people, including myself, 20 years ago in the simple bacteria system, … we did full statistical models. The systems had essential components.
There is a cultural disconnect — a lot of people came into this field with no background in biology. We did a survey [here at the conference] and only one person out of about 50 people had read Darwin.
My students are required to read two books: Darwin’s [The Origin of the Species] and [Douglas Adams’] Hitchhiker’s Guide to the Galaxy. The computer scientists are so specialized sometimes … they can’t tell you anything about Darwin, Newton, Einstein …
I am pretty close to tone deaf but I could probably list a couple classical music [pieces, off the top of my head.]
So the young people are not well rounded?
I think the way science often progresses is [that] something done in an entirely different field gets you to think of [the main scientific discipline being studied] in a different way.
I am not a biologist, [for example], I am a physicist — [but] the problems I work on are [those in] biology.
If a mathematician or computer scientist comes to me, [they should] say, ‘Yes, I am a computer scientist, but all the problems I work on are [those in] biology.’ Don’t say they are bioinformatics [problems].