In the living cell, information is carried by molecules. The outside environment and the biochemical circuitry of the cell churn out fluxes of molecular "words" that are read, processed and stored in memory by other molecules. The cell's information-processing networks often need to translate a word written in one species of molecules into another word written in a different molecular language. This requires a molecular code-table that translates between the two molecular languages. Perhaps the best-known example is the genetic code-table that translates the 64 codons, three-letter words written in DNA bases,
into the 20 amino-acids. We may think of such a code-table as a mapping between the space of molecular symbols, e.g. the codons, and the space of molecular
meanings, e.g. amino-acids.
Evolution poses the organism with a semantic challenge: its code-tables must assign symbols to meanings in a manner that minimizes the impact of the molecular recognition errors, i.e. the error-load, while keeping down the cost of resources that the code-table necessitates. Our work tries to explain how these conflicting needs drive the emergence and evolution of molecular codes. The model suggests unifying principles for diverse biological codes of varying level of complexity, such as the genetic code, the transcription regulatory network and the operons, the logical elements of the this network.
It appears from our study that evolutionary optimization with respect to noise dominates the organization and evolution of molecular codes.
Moreover, we suggest that molecular codes may emerge as a consequence of the optimization process.
Our theory describes molecular codes in terms of noisy information channels. For example, in the case of the genetic code, the amino-acids reside in an abstract 'chemistry' space whose points are all possible combinations of chemical properties. The codons reside in a space of their own. The noisy channel consists of three stages: Encoder that writes amino-acids as codons, a Reader that retrieves codons from memory and a Decoder that maps back the codon into an amino acid. All three stages are error-prone and are therefore described by stochastic matrices. The chemical dissimilarity between the original amino-acid and the final one determines the quality of the code.
Physics of Proteins
Almost all living information channels rely on the ability of bio-molecules to specifically recognize each other. The participant molecules must have the ability to interact preferentially with a specific target molecule among a vast variety of different, but structurally similar molecules. The forces that govern molecular binding have typical energies that are not much above the thermal energy, and thermal fluctuations therefore play a significant role. This motivates us to study the physical mechanisms that assist the organism to efficiently perform molecular recognition in a noisy environment.
The question of whether conformational changes, especially the induced-fit mechanism, can provide or perhaps enhance specificity, that is the ability to discriminate between competing targets, has been a matter of debate over the years. This question has been the focal point of a general discussion seeking for characteristics of molecules that could achieve optimal recognition. We study this question by introducing a statistical mechanics model that aims to elucidate the underlying relations between flexibility and conformation and the quality of molecular recognition. Another approach we use is the formulation of the molecular recognition process in terms of a Bayesian signal detection problem. Thus, we introduce a comprehensive formalism for the design of optimal bio-recognizer. We employed this framework to examine the process of homologous recmobination and the Rubisco enzyme, the most abundant protein on Earth.
Molecular binding as a detection problem.
(a) Typical molecular recognition reaction. A recognizer a can bind to two competing molecules A and B and thus initiate correct and incorrect actions. The reaction depends on the dissociation constants KAB and on the production rates AB (see text). (b) The biological recognition system can be regarded as a detection problem where both the input-output signals and the decision unit are molecules. On the molecular level, the decision is carried out through binding of the recognizer a to the ’input’ molecules, A and B. The molecular binding, which is governed by the physical properties of the interacting molecules, dictates the decision quality.
Evolution Theory
Many organisms in nature are known to sacrifice themselves for the benefit of others.
This phenomenon, which is known as altruism, and may seem to contradict natural selection,
is in fact a consequence of selection at the gene level. We investigate altruism and its consequences
using tools of evolutionary dynamics and evolutionary game theory.
We find that altruism is a broad characteristic that
may include a diversity of phenomena that do not appear to be altruistic at first glance,
for example, mutation and antibiotic resistance. We suggest that the tendency
toward altruism is stronger when the population is small and possibly subject to environmental
pressures. Therefore, some organisms may choose to behave altruistically only
when subjected to certain conditions. We examine how that this may essentially
change these organisms’ evolutionary dynamics, and enhance the stability and diversity
of biological and ecological systems. We plan to study the use of the emergent
altruistic dynamics as a possible mechanism to optimize evolutionary algorithms.
Living Neural Networks (collaboration with Elisha Moses)
Neurons form complex webs of connections. Dendrites and axons extend, ramify, and form synaptic links with their neighbors. This complex wiring diagram has caught the attention of physicists and biologists for its resemblance to problems of percolation. Important questions are the critical distance that dendrites and axons have to travel in order to make the network percolate, i.e., to establish a path from one neuron of the network to any other, or the number of bonds (connections) or sites (cell bodies) that can be removed without critically damaging the functionality of the circuit.
In the brain, neural networks display such robust flexibility that circuits tolerate the destruction of many neurons or connections while keeping the same, though degraded, function. For example, it is currently believed that in Parkinson's disease, up to 70% of the functionality of the neurons in the affected areas can be lost before behavioral symptoms appear. At the core of the experiments and the model is a percolation approach. In our work, we consider a simplified model of a neural network in terms of bond-percolation on a graph.
A graph representation of a 2D neural network used in our percolation approach. The vertices of the graph are the neurons and dendrites or axons are denoted by edges. In the experiment, the average connectivity c is varied by adding reagents that control the activity of the synapses. The percolating network responds by changing the fraction of the giant component g .
Physics of Microfluidics (collaboration with Roy Bar-Ziv)
The emergence and propagation of sound waves in a crystal shed light on the fundamental forces between the particles that make up the crystal. Periodic arrays of rapidly flowing droplets arise naturally in microfluidic devices, which have recently gained much attention as promising technological tools. Microfluidic devices operate at the low Reynolds number regime where inertia is negligible with respect to viscous dissipation. It is therefore expected that vibrations in droplet crystals will be strongly over-damped just like the typical vibrations of micro-particle colloidal crystals that belong to the same viscosity-dominated regime.
We have recently observed of acoustic waves in a one-dimensional crystal of water-in-oil droplets flowing in a microfluidic device. The dispersion relations of these waves have unique properties markedly different from those of a harmonic crystal. In addition, these vibrations sometime lead to crystal instabilities. We have shown theoretically how the symmetry-breaking flow field induces the underlying hydrodynamic inter-droplet forces that govern these 1D crystals, which result in waves and instabilities. This has obvious practical implications for the design of high-throughput microfluidic devices. Conceptually, our study places the microfluidic droplet generators in the physical context of dissipative systems that exhibit instabilities and waves, such as elastic turbulence and several driven systems.
![[Cover]](papers/MolCellCover.jpg){kind=link}