Maria Ermolaeva Interviewed by David Bradley
Posted April 13, 2001 · Issue 100

Born	February 10, 1973, Moscow, Russia
Position	Staff Scientist, the Institute for Genomic Research, Rockville, Maryland
Biography	1990-1995: Diploma with honors, Biophysics, Moscow State University, Russia. 1995-1997: Ph.D., physics and mathematics, Moscow State University. 1997-1999 postdoc, computational biochemistry, Pennsylvania State University. 1999-present: Staff Scientist, the Institute for Genomic Research.
Research interests	Computational genomics (bioinformatics). Her current research focuses mainly on developing computational methods to find or predict different structures (operons, transcription terminators, chromosome duplications, genes, ribosome binding sites, etc.) in complete genomes. Among the algorithms she has developed is a statistical analysis that estimates the accuracy of each prediction. For example, the TransTerm program finds rho-independent transcription terminators in bacterial genomes.

What was the key event that pushed you into research?

My journey into science started from mathematics. I had a wonderful math teacher [in] high school. He showed us the beauty of mathematics and logic. Later, I understood that math is not only interesting in itself, but it is also the language of nature. Math allows us to communicate with nature in its own language, ask it questions and even, sometimes, get answers. Modern physics is based on such "communication" - all physical laws are described by mathematical equations. Living systems are much more complex than physical objects, and for centuries they were only described using human language. The latest progress in biology (discovery of DNA, genes, etc.) has made it possible to apply mathematics to biology. So I decided to focus my research on applying mathematics (and computer science) to solving the mystery of life.

Which research paper has had the most effect on your work?

There are so many such papers that I could not select just one. The only possible exception is Darwin's On the Origin of Species. [It] gives a complete (for that time) picture of evolution, including natural selection. But one might ask, Who was the inventor of the natural selection concept? Darwin had published nothing concerning natural selection by the time he received Alfred Russel Wallace's letter with his thoughts about natural selection, and there is an opinion that Wallace was the first to suggest it. But I would leave this question to historians.

What was your best experiment?

My work is mostly theoretical.

In that case, what has been your most satisfying conjecture?

I usually design my work in such a way that the experimental data would already be available to check my theoretical hypothesis. My latest work relates to predicting operons [clusters of genes encoding enzymes in a particular biochemical pathway, plus their associated regulatory elements] in bacterial genomes. In order to verify my results against experimental ones, I needed a large set of experimentally documented operons. Detecting operons experimentally is not an easy task, and usually one research paper describes just one (or a few) operons. So I would need to read hundreds of papers to collect enough experimental evidence.

Fortunately, I have found two databases available on the Web that contain all this information already compiled and available to download. One of them is RegulonDB by the Collado-Vides group [at the Universidad Nacional Autonoma de Mexico] and the other one was the CGSC DB by Mary Berlyn [at Yale].

Which scientific idea do you regret the most?

In my experience, about half of scientific ideas do not work, and I deeply regret all of them. One particular idea was modeling HIV protease inhibitors (i.e., potential AIDS drugs) using semi-empirical quantum chemistry methods. Computational quantum chemistry is a very powerful tool for analyzing small molecules, but large systems could not be treated using this method due to the enormous requirements for computational time and memory. So I applied semi-empirical methods that are much less accurate, but they sometimes do produce realistic results. For the first two protease inhibitors I received results very close to experimental. This seemed to indicate that the method was working, but then the rest of the results were completely different from the experimental ones.

What qualities do you need to be a successful researcher?

I think that the main quality that a successful researcher needs is an interest in nature, a desire to understand its laws. The biological sciences, however, will soon demand one additional quality - good ethical judgment. In the near future we will reach the point when biologists will be able to do almost everything that we can now see in science fiction movies - clone people, erase one's memory and create a new one, etcetera.

Do you think it is a scientist's place to make ethical judgments?

I think that ethical decisions should, of course, be made by all people, not by scientists alone. At the same time, science is developing so fast that new laws may not be adopted quickly enough, and it may be impossible to provide a law for each possible situation. So scientists will always have to make some ethical decisions.

If you could work with any scientist - living or historical - who would it be?

It would be Werner Heisenberg, one of the authors of quantum physics. Of course, quantum physics was not related to biology (at that time), but I think that Heisenberg's scientific philosophy now can be applied to biology. His philosophy allows one to see things from a completely different point of view, which can lead to new discoveries. For example, any object (for example, a tree) consists of a large number of elementary particles (electrons, protons, etcetera). Heisenberg considered that these particles do not exist as real objects; they are just mathematical equations. In other words, any object (such as a tree) is just a large set of mathematical equations. It may have been this conceptual way of seeing things that resulted in Heisenberg's discovery of the uncertainty principle and other laws of quantum physics. I think that modern biology has now reached the point (or will reach it rather soon) when such a level of abstraction will be necessary.

So - alluding to the philosopher's query - does a falling tree make a sound if no one is there to hear it?

According to Heisenberg's philosophy, a falling tree (which is just a set of mathematical equations) produces a new set of equations (i.e., a sound) even if there is nobody in the forest.

What is your greatest unanswered scientific question?

My greatest unanswered scientific question is whether the origination and evolution of life was a random event or . . . predictable. In other words, if we could start all this from the beginning, how much time will it take for life to appear? Will it be the same form of life? Will evolution go the same way toward a human?

What is your hunch?

First, life might have been based on other molecules: the DNA/RNA/protein trinity is just a way to store and reproduce information; there could be other chemical structures that might successfully perform the same task. However, if evolution were to manage to go far enough, then it should inevitably lead to an intelligent form of life. Intelligence is the most useful adaptation; it enhances the ability of the species to survive, expands the inhabitable territory, and helps in finding new sources of food.

What scientific plans do you have for the next five years?

I am working at the Institute for Genomic Research (TIGR), one of the world leaders in genome sequencing. During the last few years an enormous amount of sequence data has been produced by TIGR and other sequencing centers, and even more data is expected in the near future. I want to apply statistical analysis to this data in order to study the evolution of life.

David Bradley, a freelance science writer, lives on the edge of the fens north of Cambridge, United Kingdom. Elemental Discoveries is his Webzine of science news.

Tell us what you think.

Gary Siuzdak

Steven Chu

Ormond MacDougald

Douglas H. Erwin

Egbert Giles Leigh

Nenad Ban