not a keynote , but a footnote on molecular biology and computation for Rocky 1

not a keynote, but a footnote on molecular biology and computation

for Rocky 1

The Biology of Information

Walter Fontana (SFI)[email protected]

www.santafe.edu/~walter

1. What can computation do for biology?

The computer as…

The computer as…

…theater: simulation, modeling

The computer as…

…theater: simulation, modeling

…library: organization of data

The computer as…

…theater: simulation, modeling

…library: organization of data

…instrument: component of experiment

The computer as…

…theater: simulation, modeling

…library: organization of data

…instrument: component of experiment

…mathematical structure: formalism, concept

. . . ( , ) ( )u eu e x e x x

1. What can computation do for biology?

1. What can computation do for biology?

Nothing.

1. What can computation do for biology?

A lot.

1. What can computation do for biology?

2. What can biology do for computation?

…but this business is not well understood on both sides…

molecular biology and computer science are in the same conceptual business

molecular biology and computer science are in the same conceptual business

at the very minimum,both are about structure-behavior relations,

i.e. configuring systems to engender specific behaviors(both are “programming” disciplines)

a self-printing program in C

a self-printing program in C

now imagine these expressions…

… decaying… moving around… combining into imprecise meanings… acting in parallel & asynchronously

a self-printing program

now imagine these expressions…

… decaying… moving around… combining into imprecise meanings… acting in parallel & asynchronously

molecular components…

…turn over (from minutes to days)…are stochastic (wrt reliability, number, recognition)…move around (passively or actively) in a structured medium…communicate through physical contact…control each other’s state and production…are often multipurpose…need (lots of) energy for communication…operate concurrently

turn-over of components:persistence of identitymemory of state

stochasticity (in number and recognition):error-correction

massive concurrency:emergence of determinismcoordination & conflicts

communication by contact:energy transportcontrol of space

…which entails a suite of issues, such as:

plasticityreconfigurabilitycompressibilityevolvability (neutrality, modularity)autonomyselfrobustness

biological architectures emphasize systemic capacities, e.g.

all these features are desirable but absent in present daycomputer architectures

+

in biological systems, there is no “software running on something” !

IS NOT

in (theoretical) computer science…

…physical hardware is distinct from software.(in CS, “machine” is a software notion)

in biology…

…physical hardware is software

• dynamics

• stochasticity

• effective potentials

• combinatorial trajectories & path-dependency

• discrete events & concurrency

• object syntax and action

• generative interactions

physics

logic

digital

analog

A few vignettes where the gap between computation and molecular biology is widest

enzyme kinetics 101

Who is the “s

ignal”??

phosphorylation chain

phosphorylation chain

multiple phosphorylation in proteins (phosphobase*)

* A. Kreegipuu, N. Blom, S. Brunak. Nucleic Acids Research (1998/1999)

W.Fontana & D.Krakauer (in progress)

0

1

( / )J i

iQ S

phosphorylation chain and hypersensitivity

generalized signaling cascades

shifting the threshold by positioning P-chains of different width at various depths in a cascade

pulse filter

multiple phosphorylation as pulse filter

W.Fontana & D.Krakauer (in progress)

multiple phosphorylation as pulse filter

W.Fontana & D.Krakauer (in preparation)

memory and “checkpoints”

phosphorylation chain

phosphorylation chain with positive feedback

phosphorylation chain with symmetric feedback

phosphorylation chain with symmetric feedback

|rela

tive a

vera

ge d

iff o

f end s

tate

s|

n/signal

large J:Bose-Einstein

small J:Curie-Weiss

S.Krishnamurty,E.Smith,D.Krakauer,W.Fontana

Phys.Rev.Lett., submitted

stochastic treatment of a P-chain with symmetric feedback

second order phase-transition

stochastic master equation

introduce operator algebra familiar from many-body physics

obtain equivalent equation,now approachable by techniques

from many-body physics

effective potentials

idea by M.Sasai & P.Wolynes:

Sasai & Wolynes: “Stochastic gene expression as a many-body problem”,PNAS, 100, 2374–2379 (2003).

the landscape concept made formally preciseby techniques from statistical mechanics

“programming” becomes sculpting an appropriate landscape.

But how?(cf. neural networks, spin glasses…)

the landscape metaphor: from energy landscapes in proteins to epigenetic landscapes a la Waddington

reconfigurable molecular networks, plasticity

Milan N Stojanovic, Darko Stefanovic. Nature Biotechnology, 21, 1069 - 1074 (2003)

allostericRNA gates

Why do we need the formalisms of computation and logic?

a pragmatic answer: more tools get us to more places.

a deeper answer: because we need a theory of (molecular) objects.Why?

Because the pressing (and recalcitrant) question for biology is not only to describe the behavior of a particular system, but to understand that system in the context of the possible, i.e. of what is evolutionarily

accessible to it.

Stated differently: we must eventually be able to reason about novelty.We never can do so within the confines of dynamical systems,

because dynamical systems do not represent the objects they are made of.(Remember chemistry.)

we need an abstraction of chemistryin which

molecules are interacting computational agents

the grand challenge:

describe a system with an expression that is at the same time

a program to “run” that systemAND

a formula to reason about it abstractly.

A brief coda where the gap between computation and molecular biology is closing

(at the formal language end)

inputoutput

function

no interaction with the “environment”

Old notion of computation

semantics: input-output relation

process

semantics: potential sequences of interaction events

interaction with the “environment”

New notion of computation

function

closed system

process

open system

computation:

analogy in physics:

equilibrium normal form

organizationmain concern:

Theory of concurrency, Process algebra

Robin Milner, Communicating and Mobile Systems: the -calculus, Cambridge (1999)

The -calculus (Milner, Walker and Parrow 1989)

• a program specifies a network of interacting processes

• processes are defined by their potential communication activities

• communication occurs on complementary channels, identified by names

• message content: channel name

Aviv Regev, Ehud Shapiro, Corrado Priami, and others:application of concurrency / process algebras

to molecular signal transduction

A.Regev & E.Shapiro, Nature, 419, 343 (2000), Concepts

concurrency theory, what for?

• tool for agent-based simulation based on a theory of the agents

• tool for agent-based simulation

at worst:

at its most hopeful:

molecular biology

nanotechnology &molecular information systems

distributed OS design

concurrency

a lingua franca?