Method of generating an optimized, diverse population of variants

• US 8,504,498 B2
• Filed: 02/12/2009
• Issued: 08/06/2013
• Est. Priority Date: 02/12/2008
• Status: Active Grant

• Current Assignee: Codexis, Inc.
• Original Assignee: Codexis, Inc.
• Inventors: Fox, Richard
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): RIOJAS RAMIREZ, MARIO
• Application Number:
  US12/867,429
• Publication Number:
  US 20110029468A1
• Time in Patent Office: 1,636 Days
• Field of Search: None
• US Class Current: 706/13
• CPC Class Codes: G06N 3/126 Evolutionary algorithms, e.g. genetic algorithms or genetic programming
• CPC Class Codes: G16B 35/00 ICT specially adapted for in silico combinatorial libraries of nucleic acids, proteins or peptides
• CPC Class Codes: G16B 35/20 Screening of libraries
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• CPC Class Codes: G16C 20/60 In silico combinatorial chemistry

Abstract

The present disclosure relates to methods of rapidly and efficiently searching biologically-related data space to identify a population set maximally diverse and optimized for sets of desired properties. More specifically, the disclosure provides methods of identifying a diverse, evolutionary separated bio-molecules with desired properties from complex bio-molecule libraries. The disclosure additionally provides digital systems and software for performing these methods.

Claims

1. A computer program product comprising a non-transitory computer useable medium having a computer readable medium having program instructions for selecting an optimized, diverse set of molecular variants from a plurality of molecular variants, wherein the program instructions comprise:

(a) code for receiving an objective data set representing two or more properties of each molecular variant in a plurality of molecular variants;

(b) code for generating a first Pareto front membership of all molecular variants based on the one or more objectives for optimization;

(c) code for setting optimization parameters, wherein the optimization parameters comprise;

(i) number nvar of molecular variants to create;

(ii) a molecular variant population size popSize;

(iii) a crossover rate crossrate;

(iv) a mutation rate mutrate;

(v) a fitness function comprising a penalty fitness function and an overall fitness function, wherein the penalty fitness is based on niche count and the overall fitness function is based on location in a descending Pareto front divided by the shared number of molecular variants within each front; and

(vi) specific molecular variants to be included in the nvar of molecular variants; and

(d) code for identifying a search space of acceptable molecular variants;

(e) code for generating a random population of genomes from the search space of acceptable molecular variants by applying a selection operator; and

(f) code for applying a set of operators comprising a crossover operator, a mutation operator, a fitness operator, and a repair operator on the random plurality of genomes to select the nvar set of optimized, diverse molecular variants.

2. The computer program of claim 1 further comprising a code for repeating step (f) on a combined population of the first nvar set molecular variant and the molecular variant set popSize to generate a second nvar set of molecular variants.

3. A computer program product comprising a non-transitory computer useable medium a computer readable medium having program instructions for selecting an optimized, diverse set of molecular variants from a plurality of molecular variants, wherein the program instructions comprise:

(a) code for receiving an objective data set representing two or more properties of each molecular variant in a plurality of molecular variants;

(b) code for generating a first Pareto front membership of all molecular variants based on the one or more objectives for optimization;

(c) code for setting optimization parameters, wherein the optimization parameters comprise;

(i) number nvar of molecular variants to create;

(ii) a molecular variant population size popSize;

(iii) a crossover rate crossrate;

(iv) number of generations to create nGen;

(v) a mutation rate mutrate;

(vi) a fitness function comprising a penalty fitness function and an overall fitness function, wherein the penalty fitness is based on niche count and the overall fitness function is based on the location of the molecular variant in a descending Pareto front divided by the shared number of molecular variants within each front; and

(vii) specific molecular variants to be included in the nvar of molecular variants; and

(d) code for identifying a search space of acceptable molecular variants;

(e) code for generating a random population of genomes from the search space of acceptable molecular variants;

(f) code for selecting a first set of genomes of size popSize from the random population, wherein each genome consisting of nvar molecular variants is created by applying a selection operator, a crossover operator, a mutation operator, a repair operator, and a fitness operator to the random population of genomes; and

(g) code for returning the genome with the highest fitness as the final, optimized, diverse nvar set of molecular variants.

4. The computer program of claim 3 further comprising a code for repeating step (f) nGen times to select a new population of genomes of size popSize.