Herein lies madness

[Building | Usage | Issues | Download | Links]

HPFold implements an Ant Colony Optimisation (ACO) algorithm for the 2D hydrophobic polar protein folding problem. It is based on an article by Alena Shmygelska and Holger H Hoos, see Links section below.

Building HPFold

HPFold is written in C, and should compile on any Unix system. Unpack the source code (tar xvfz hpfold-X.Y.tar.gz), change directory to the hpfold-X.Y/ directory and run make. A hpfold executable will be built in the bin/ subdirectory.

Usage

HPFold allows for configuration of a number settings, ex.:

# hpfold --help
Usage: hpfold [OPTION]... <SEQ>
Fold SEQ.

  -r, --runs=NUM                number of runs (default: 10)
  -i, --iterations=NUM          number of iterations (default: 50000)
  -a, --ants=NUM                number of ants (default: 100)
  -o, --optimal=NUM             optimal (or approximated) energy
  -l, --improving-ants=NUM      fraction of ants performing local search
  -e, --elitist-ants=NUM        fraction of ants updating pheromone trails
  -s, --search-length=NUM       maximum number of local search iterations
                                  not producing improvements (default: 1000)
  -c, --cpu-seconds=NUM         maximum number of CPU seconds per run.
  -v, --verbose                 increase output verbosity.
      --alpha=NUM               pheromone importance (default: 1.0)
      --beta=NUM                heuristics importance (default: 2.0)
      --rho=NUM                 pheromone persistence (default: 0.8)
      --theta=NUM               pheromone threshold (default: 0.05)
      --version                 display version information and exit
      --help                    display this help and exit

However, the default settings are reasonably sensible, and only the optimal (or approximated) energy -o as well as a sequence must be provided. Using default settings will result in 10 runs using a maximum of 120 CPU seconds per run, ex.:

# hpfold -o -21 hhphphphphhhhphppphppphpppphppphppphphhhhphphphphh
Parameter settings:
===================
number of runs = 10
number of iterations = 50000
number of ants = 100
optimal energy = -21
maximum CPU seconds per run = 120

local search length = 1000
local search ant percentage = 0.500000
elitists percentage = 0.010000

pheromone importance (alpha) = 1.000000
heuristics importance (beta) = 2.000000
pheromone persistence (rho) = 0.800000
pheromone threshold (theta) = 0.050000

sequence = hhphphphphhhhphppphppphpppphppphppphphhhhphphphphh
sequence length = 50

24 hydrophobes (8 hydrophobe contacts)

run = 1; iterations = 2; energy = -21; fold = frrfrfrlrrlrllrllfrfllrfffffflfrllfrllflrrfrlrfrr; cpu = 2.276
run = 2; iterations = 6; energy = -21; fold = frrfrlrfrrlffrllffffffrfllfffllrffffrfrrllrlflfll; cpu = 8.813
run = 3; iterations = 3; energy = -21; fold = frrfrlrfrrllrrlfrrllrflfrrflfrflrrflrrllrrfrlrfrr; cpu = 3.200
run = 4; iterations = 2; energy = -21; fold = frrfrlrfrrlfllrfllrflfffffffflfrllfrllrrllflrlfll; cpu = 1.900
run = 5; iterations = 3; energy = -21; fold = fllflflrllrlrrlffflfrrlfrrflfrrllrfrlfllrrlrfrfrr; cpu = 4.436
run = 6; iterations = 2; energy = -21; fold = fllflrlfllrrllrfllrflfffrrffflfrllfrllflrrfrlrfrr; cpu = 2.668
run = 7; iterations = 2; energy = -21; fold = frrfrlrfrrllrrlfrrllrflfrrflfrflrrflrrfrllflrlfll; cpu = 1.892
run = 8; iterations = 2; energy = -21; fold = fllflrlfllrrllrfllrflfrfllfrlrfrllfrllflrrfrlrfrr; cpu = 2.576
run = 9; iterations = 14; energy = -21; fold = frrfrlrfrrllrrlfrrllrflfrrflfrflrrflrrfrllflrlfll; cpu = 19.821
run = 10; iterations = 2; energy = -21; fold = frrfrlrfrrllrrllrfrfllrfllfrfllrfrllrrfrllflrlfll; cpu = 1.692

Known issues

Deadlocks may (rarely) occur during the backtrack phase. This is currently solved by skipping ants where the number of backtracking attempts exceed the sequence length.
Code could use a cleaning.

Download source code

v0.1.2