Research
Bioinformatics

 

Introduction

HYDRA for (HYbrid Distance Recombination Algorithm) is a phylogenetic reconstruction software based on the maximum parsimony criterion under Fitch's parsimony. The core of Hydra relies on a memetic algorithm, i.e. a combination of a Genetic Algorithm (GA) and a Local Search (LS) algorithm.

The main features of Hydra are the introduction of a specific crossover operator for trees called DiBIP (Goeffon et al, 2006) used with the GA and the definition of a new neighborhood called progressive neighborhood (Goeffon et al, 2007) used for LS. They both rely on the notion of topological distance.

HYDRA is provided with a set of benchmarks from C. C. Ribeiro. The current version 1.1 is more efficient than version 1.0 when compiled on an Intel or AMD processor implementing SSE2 instructions.


Download

Hydra is freely available for non commercial purpose.



Some results

ProblemHydraAG+PRTNT
tst01545549547
tst02135413581361
tst03833838840
tst04588592595
tst05789790789
tst06596603601
tst07127012761272
tst08853863866
tst09114511501146
tst10721725721
best results from random instances of C. C. Ribeiro

Bibliography

Adrien Goeffon, Jean-Michel Richer, Jin-Kao Hao, Progressive Tree Neighborhood applied to the Maximum Parsimony Problem, In IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB), 2007 (to appear pdf).
Adrien Goeffon, Jean-Michel Richer, Jin-Kao Hao, A Distance-based Information Preservation Tree Crossover for the Maximum Parsimony Problem, In Proceedings of Parallel Problem Solving From Nature IX, (PPSN 06), Reykjavik, September 2006 (pdf).
marqueur eStat\'Perso