Raj Chakrabarti @ home

Since the early days of evolutionary theory, the remarkable organization in the biological world has evinced a mixture of awe and wonder. One of the most seminal moments in the history of science was the discovery in the late 19th century that this order is not created by design, but rather originates due to the algorithmic principle of random genetic variation and natural selection.  The question as to whether additional -physical- organizing principles also play a role in the generation of this order, or whether random variation and selection alone is sufficient to produce biological complexity, has remained largely unanswered, since it is notoriously difficult to prove.  This debate originated in the differing views of Alfred Russell Wallace and Charles Darwin in the 1850's. 

Wallace appeared to have envisioned natural selection as a kind of feedback mechanism keeping species and varieties adapted to their environment. In a well-known speech before the Linnean Society in 1858, Wallace stated:


However, there was no evidence for the conjecture that feedback control principles were exploited in the evolution of organisms. Many modern-day authors such as Daniel Dennett, e.g., contend that the algorithm of natural selection, without additional self-organizing principles, are sufficient for generating all the complexity in the living world around us. However, many notable evolutionary biologists, including Steven J. Gould, have contested this claim. Among biophysicists, Stuart Kauffman has in particular been a major proponent of the thesis that self-organizing principles based on statistical mechanics are essential for the origin and refinement of life.

In the latter half of the 20th century and into the 21st, more quantitative techniques have been deployed in order to interrogate the evolutionary dynamics of biological systems. An important milestone was the introduction of the quasispecies theory for evolutionary dynamics by Eigen and Schuster in the 1970s. This framed natural selection in terms of of self-replicating families of biopolymers, enabling the application of more sophisticated tools of dynamical systems theory to evolution.

---

My research in this area employs these tools from dynamical systems theory, along with experimental mutagenic evidence and statistical estimation techniques, to quantitatively probe questions of biological self-organization at a molecular level. 

The key to our approach is that it remains fully consistent with evolutionary theory while pointing to the ability of natural selection to exploit powerful organizing steering principles that maximize evolutionary fitness. 

Below are some representative papers that illustrate our approach.



Evidence for the impact of self-organizing principles of control and systems theory to biological evolution has been virtually nonexistent since the time of Wallace and Darwin. We propose  a new evolutionary theory, based on experimental data, which extends the standard Darwinian model of fitness maximization by incorporating the principles of optimal control.

Reprinted in: Virt. J. Biol. Phys. 16: 1, July 2008.

Princeton University has drafted a press release on this work.

---

Optimal control of evolutionary dynamics

2.   R. Chakrabarti, H. Rabitz, and G. L. McLendon. Optimal control of evolutionary dynamics. arXiv:0806.2331v1 [quant-bio] (2008).

The quasispecies evolutionary dynamics model in the above paper, and the generality of control phenomena in evolution, are examined in greater detail.

---

Computational prediction of native protein-ligand binding and enzyme active site sequences

3.  R. Chakrabarti, A. M. Klibanov, and R. A. Friesner, Proc. Natl. Acad. Sci. USA 102: 10153-10158 (2005). 
   
     Supporting material

We show, for the first time, that it is possible to predict the identities as well as the
conformations of nearly all the functional surface amino acids in proteins based purely
on physical models and numerical search algorithms

Highlighted in:  "Progress in computational protein design", Curr. Opin. Biotechnol. 18: 1-7 (2007).

---

Sequence optimization and designability of enzyme active sites

4.  R. Chakrabarti, A. M. Klibanov, and R. A. Friesner, Proc. Natl. Acad. Sci. USA 102: 12035-12040 (2005).    
     Supporting material

The first quantitative physical investigation of the fitness measures underlying the natural evolution of biocatalysts.

Highlighted in: "Do-it yourself enzymes", Nature Chemical Biology 4, 273 - 275 (2008)

Papers (3) and (4) were recently highlighted in the biophysics literature due to their efforts to delineate fundamental principles of computational enzyme engineering.

---

Next steps

In future work, I will continue to study the evolutionary dynamics of enzyme evolution. An important question raised by (3) and (4) is whether enzyme evolution is a multiobjective optimization problem, where tradeoffs between physical properties including protein stability, substrate binding, catalytic activity and product release generate a "Pareto front" of nondominated solutions. Multiobjective evolutionary algorithms (MOEAs) will be used to optimize active site sequences of natural enzymes, as well as to design artificial enzymes,  in order to test this hypothesis.

The central idea here is to simulate the evolution of the active site by assigning variable weights to these various physical properties and then to determine (using rigorous statistical methods) which weights (or range of weights) are necessary to faithfully reproduce the sequence distributions observed in natural enzyme families. The fact that the organismic environment of enzymes might place different pressures on active sites from different species could necessitate the use of a range of weights, rather than a single set of weights; e.g., tighter substrate binding but more facile product release may be required in some species than in others, for optimal organismic evolution. The key point is that no matter what the organismic environment, evolution of the enzyme family can always be modeled in terms of some range of weighted combinations of these physical properties of the active site, using the biophysical techniques we have pioneered.

Further technical details on computational methodologies under development for sampling protein sequence and conformational space, as well as modeling reactive chemistry, can be found in the research proposal "High-resolution enzyme design". Applications to the problem of cellulase design - for the production of next-generation cellulosic biofuels in response to the nation's energy crisis - are outlined in "Cellulosic ethanol".

     


 

---

Optimization and complexity theory

---

Biophysics

Theoretical biophysics

Biomolecular reaction
engineering