ACO algorithms for the Hydrophobic Polar Protein Folding Problem version 3 (ACO-HPPFP-3)
===============================================================
PARAMETERS:
The following parameters NEED to be specified (no default values):
parameter description
--------------------------------
-seq hp sequence to be folded in lower case letters (e.g. hhhhpppphhh)
-eng optimal energy (specify as a positive value) known or approximated (if not known use large enough number)
-file output file
Optional parameters that have default values (same default values for 2D and 3D):
parameter default value description
-------------------------------------------------------------
-runs 10 number of independent runs of an algorithm
-tries 1000000 cut-off number for number of iterations (construction-local search-pheromone update) in each run
-time 1000000 cut-off CPU time [sec] per run
-ants 100 number of ants in the colony
-noImprLS 10000 length of local search - number of non improving steps tolerated in each call to local search
-alpha 1 weight for the pheromone information
-beta 2 weight for the heuristic information
-rho 0.8 pheromone persistence
-noise 0.05 theta (pheromone noise)
-elitist_ants 0.01 fraction of elitist ants (updating pheromone matrix, 0 <= elitist_ants <= 1)
-fractLSants 0.5 fraction of ants chosen to perform local search (0 <= fractLSants <= 1)
-probPrev 0.5 probability of keeping previous direction (in long-range move)
=================================================================
EXAMPLES OF USAGE:
folding in 2D:
./antsystem2D -seq pphpphhpphhppppphhhhhhhhhhpppppphhpphhpphpphhhhh -eng 23 -file seq1_3D -noImprLS 1000 -noise 0.05 -tries 1000000 -runs 3 -ants 100 -alpha 1 -beta 2 -rho 0.8 -elitist_ants 0.01 -fractLSants 0.5 -probPrev 0.5 -time 172800
folding in 3D:
./antsystem3D -seq hphhpphhhhphhhpphhpphphhhphphhpphhppphpppppppphh -eng 32 -file seq1_2D -noImprLS 10000 -noise 0.05 -tries 1000000 -runs 3 -ants 100 -alpha 1 -beta 2 -rho 0.8 -elitist_ants 0.01 -fractLSants 0.5 -probPrev 0.5 -time 172800
For more detailed information about the parameter settings both in 2D and 3D
check the paper: An ant colony optimisation algorithm for the 2D and 3D hydrophobic polar protein folding problem. Alena Shmygelska and Holger H Hoos,BMC Bioinformatics 2005, 6:30 (14 Feb 2005).
==================================================================
OUTPUT:
Console output produced:
ACO algorithms for the Hydrophobic Polar Protein Folding Problem version 3 (ACO-HPPFP-3)
run 0
run 1
run 2
...
[Algorithm display progress in terms of number of independent runs completed]
==================================================================
File output produced:
ACO algorithms for the Hydrophobic Polar Protein Folding Problem version 3 (ACO-HPPFP-3)
Parameter-settings:
max runs = 3
max iterations = 1000000
cut-off time = 172800 CPU sec
optimal energy (known or approximated) = -32
number of ants = 100
length of local search - noImpr steps = 10000
alpha (pheromone weight) = 1
beta (heuristic info weight) = 2
rho (pheromone persistence) = 0.8
theta (pheromone noise) = 0.05
percent of elitist ants (updating pheromone matrix) = 1 %
percent of ants chosen to perform local search = 50 %
probability of keeping previous direction (in long-range move) = 0.5
HP sequence: hphhpphhhhphhhpphhpphphhhphphhpphhppphpppppppphh
sequence size = 48
==============================================
run = 0; iteration = 1; CPU time = 0 sec; Lowest energy = -11; Conformation = udlrrduurrudrddudrudllurdldggrluruslsdlruldsduul;
run = 0; iteration = 1; CPU time = 0 sec; Lowest energy = -13; Conformation = rudrssdruddslsllurdduurrssggruurrulsdrsdslldrddu;
...
[Parameter settings are outputed; in each run for each improving energy value - Lowest energy we output: CPU time [sec] required to find this solution quality, iteration number in which this solution quality was found, and the actual conformation of a protein in terms of relative directions [s-straight; l-left; r-right; u-up; d-down; g-grounded amino acid (there is a pair of amino acids that serve as the point of reference)].
=====================================================================