SRI International Bioinformatics 1 A BRG Biofuels Metabolic Engineering Project Bioinformatics Research Group SRI International [email protected] [email protected].

SRI International Bioinformatics1

A BRG Biofuels Metabolic Engineering Project

Bioinformatics Research Group

SRI International

[email protected]

[email protected]

[email protected]

Ron CaspiMario Latendresse

Peter Karp

October 2010 Pathway Tools Workshop

SRI International Bioinformatics2

Two Main Parts

This project will be conducted in the BRG group. Funded by the Department of Energy (DOE) under the direction of Peter Karp, PI

Soon to be started The project has two main parts:

1. Curating MetaCyc for biofuel pathways under the supervision of Ron Caspi, co-PI;

2. A computational tool to infer new pathways, co-PI, Mario Latendresse. (Not biofuels specific)

SRI International Bioinformatics3

Computational Tool

Basic statement of the problem:

Given a source and target metabolites, what are the possible metabolic pathways from the source to the target in a given organism?

And rank these pathways based on their qualityAnd constrained these pathways: number of

carbon atoms from the source to the target, length of pathway, number of genes to delete or add, and so on

SRI International Bioinformatics4

Computational Tool Approach

Use a searching algorithm (A*) that can use expert knowledge as well as precise evaluation algorithms of the quality of pathways found

Based on 1. Tracing of carbon atoms in pathways. From the source to

target, the number of carbon atoms can be evaluated.2. Gibbs free energy of reactions will help the evaluation of

the feasibility of existing and new pathways. A prototype A*, based on compound similarity,

has been written to find the shortest pathway between two compounds

SRI International Bioinformatics5

The A* algorithm

A general searching algorithm in a tree-like search space

Uses two evaluating functions g(path) and h(rest-path)

Function g evaluates the path (length, Gibbs energy, cost of enzymes, and so on) seen so far

Function h evaluates the potential of current path (based on heuristics) to the target metabolite

Next node to search is based on the value of h: go in the best direction

Function h can use expert knowledge as well as precise calculations

SRI International Bioinformatics6

A* Searching

SRI International Bioinformatics7

A* Prototype Results

Found all the pathways described in the paper by McShan, Rao, Shah on Pathminer, Bioinformatics, Vol. 19, no 13, 2003.

For example, from citrate to L-histidine (E. coli)In less than a second or a few seconds for a

seven step pathwaysProof of concept since the feasibility of such

pathways is not proven: not enough constrained applied so far

SRI International Bioinformatics8

Gibbs Free Energy Computation by the Group Contribution MethodPrototype implementation of Jankowski’s methodStill needs a full evaluation

SRI International Bioinformatics9

Curation of Biofuel Pathways in MetaCyc

1. Identify lignocellulose degrading enzymes and curate them into MetaCyc

2. Expand the coverage of hydrogenases in MetaCyc to include representatives of all known classes and subtypes

3. Expand MetaCyc coverage of fatty acids and lipids biosynthetic enzymes to reflect currently known metabolic and genetic information about oil production in algae and cyanobacteria

4. We will identify biochemically verified pathways involved in processes relevant to biofuel production

SRI International Bioinformatics10

Graphical Extension of Pathway Tools

Show complex polymers degradation processes that involve multiple enzymes attacking simultaneously a complex molecule at different sites

SRI International Bioinformatics11

Validating the Computational Tool

The new biofuel pathways will be used to evaluate the capability of the computational tool.

SRI International Bioinformatics12

A BRG Biofuels Metabolic Engineering Project

Questions?