What SAT can do for BioInformatics

1
What SAT can do for BioInformatics ?
Haplotype Inference
Compact Models and Efficient Algorithms

• Problem Description Given a set G of n
  genotypes, find a set H of 2n haplotypes (not
  necessarily distinct), such that for each
  genotype gi in G there is at least one pair of
  haplotypes (hj,hk), with hk and hj in H such that
  the pair (hj,hk) explains gi
• Technology used SAT and Pseudo Boolean
  Optimization
• Reference J. Marques-Silva and I. Lynce, AAAI 06

• SAT is a well known problem and a area of
  intensive research
• SAT and SAT extensions provide compact models and
  efficient algorithms
• Efficient solvers are available for both SAT and
  SAT extensions

Checking Pedigree Consistency
SAT SAT Extensions

• Problem Description Given a pedigree check if it
  is consistent with the Mendelian laws of
  inheritance, i.e., check if every individual
  inherits exactly one allele from each of its
  parents
• Technology used SAT by translation into CNF
• Reference P. Manolios, M. Galceran Oms and S.
  Oliva Valls, TACAS 07

Protein Folding
Genome Rearrangement Distance

• Problem Description Compute the exact genomic
  distance between two genomes in the presence of
  duplications, namely, determining the maximum
  number of common intervals between the genomes
• Technology used Pseudo Boolean Optimization
• Reference S.Angibaud, G.Fertin, I.Rusu and
  S.Vialette, RECOMB-CG 06

• Problem Description Given a sequence of amino
  acids of a protein, determine the structure to
  which the protein folds
• Technology used Pseudo Boolean Optimization
  (PBO)
• Current stage currently a first attempt to
  directly translate the HP Lattice Model to PBO
  has been done
• Given the sequence
• PHPPHPHP the
• energy of the
• conformation in the
• right is -2
• Expected future steps
• Inclusion of lower bounds that are able to
  efficiently reduce the search space
• Introduction of techniques to break and remove
  symmetries

Other?

• Besides SAT itself, other extensions of SAT might
  be used to solve problems of BioInformatics
• Pseudo Boolean Optimization
• Satisfiability Module Theories
• SAT Enumeration
• Counting SAT
• Quantified Boolean Formulas

António Morgado - ajrm_at_soton.ac.uk Dr. João
Marques Silva - jpms_at_ecs.soton.ac.uk
This work is being supported by Microsoft
Research through its European PhD Scholarship
Programme