Wndchrm test harness
Jump to navigation
Jump to search
#!/usr/bin/perl
# by Chris Coletta Wed Mar 17, 2010
# The purpose of this script is to generate experimental image sets
# to perform a `wndchrm test` upon.
use strict;
use warnings;
use File::Copy;
sub RunExperiment;
if( $#ARGV < 0 ) {
print "Insufficient arguments.\n";
print "Usage: experiment [-verbose] [-dryrun] <experiment config file>\n";
print "Try again!";
die;
}
my $config_file = undef; # = "/home/colettace/data/monkeys/experimental_sets/four_way_twenty_slide/four_way_twenty_slide.experiment";
my $dryrun = 0;
my $verbose = 0;
for( my $z = 0; $z <= $#ARGV; $z++ ) {
# print "Arg $z: $ARGV[$z] \n";
if( -e $ARGV[$z] ) {
$config_file = $ARGV[$z];
}
elsif( $ARGV[$z] eq '-dryrun' ) {
$dryrun = 1;
print "Executing dry run only...\n" ;
}
elsif( $ARGV[$z] eq '-verbose' ) {
$verbose = 1;
print "Printing debug output\n";
}
else {
print "Don't recognize argument \"$ARGV[$z]\"\n";
print "Usage: experiment [-verbose] [-dryrun] <experiment config file>\n";
print "Try again!";
die;
}
}
if( !defined $config_file ) {
print "You didn't specify an experiment file..\n";
print "Usage: experiment [-verbose] [-dryrun] <experiment config file>\n";
print "Try again!";
die;
}
open( CONFIG_FILE, $config_file ) or die "Could not open experiment file: $!";
my $DEBUG1 = 1;
my $experiment_name = undef;
my $path_to_wndchrm = undef;
my $source_dir = undef;
my $target_dir = undef;
my $tile_scheme = undef;
my $large_feature_set = undef;
my $training_images = undef;
my $test_images = undef;
my $test_repetitions = undef;
my $feature_usage_fraction = undef;
my $html_file = undef;
my $write_fisher_weights = undef;
my $training_args = undef;
my $test_args = undef;
my $silence_wndchrm = undef;
my $experiment_repetitions = 1;
my $pairwise_flag = 0;
my $path_to_phylip = undef;
my $phylip_command = undef; # use a separate string for for storing phylip part of wndchrm command
# line because we need to keep track of where it is to copy dend files.
my %group_info;
my %class_config_info;
if( $DEBUG1 or $verbose or $dryrun ) {
print "Reading in config file $config_file\n";
}
while( <CONFIG_FILE> )
{
if( $DEBUG1 or $verbose or $dryrun ) {
print $_;
}
if( $_ =~ /^#/ )
{
if( $DEBUG1 or $verbose or $dryrun ) {
print "\tComment line skipped.\n";
}
next;
}
elsif( $_ =~ /^experiment\s+(\S*)/ ) {
$experiment_name = $1;
print "\tExperiment name: $experiment_name\n" if( $DEBUG1 or $verbose or $dryrun);
}
elsif( $_ =~ /^path_to_wndchrm\s+(\S*)/ ) {
$path_to_wndchrm = $1;
print "\tPath to wndchrm: $path_to_wndchrm\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^source_dir\s+(\S*)/ ) {
$source_dir = $1;
print "\tSource dir: $source_dir\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^target_dir\s+(\S*)/ ) {
$target_dir = $1;
print "\tTarget dir: $target_dir\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^class\s+(\S+)\s+(\S+)\s+(\S+)$/ )
{
if( $DEBUG1 or $verbose or $dryrun ) {
print "\t$3 images (and associated files) from directory $1 will go into class $2\n";
}
push @{ $class_config_info{$2} }, { dir_name => $1, img_quant => $3 };
}
elsif( $_ =~ /^class\s+(\S+)\s+(\S+)\s+(\S+)\s+group\s+(\S+)$/ )
{
# Sturcture of hash table %class_config_info:
# - keys %class_config_info = the name of the class in the experiment, into which
# images from multiple directories will go
# - $class_config_info{ "some key" } = reference to an array of hashes
# - @{ $class_config_info{ "some key" } } = an array of hashes containing the configuration
# info for each directory's images.
$pairwise_flag = 1; # let the script know that we'll have to do pairwise experiment
if( $DEBUG1 or $verbose or $dryrun ) {
print "\tclass $2 from dir $1: group $4, num images $3\n";
}
push @{ $class_config_info{$2} }, { dir_name => $1, img_quant => $3, group => $4 };
push @{ $group_info{$4} }, $2;
}
elsif( $_ =~ /^tile_scheme\s+(\d)/ ) {
$tile_scheme = $1;
print "\tTile scheme: $tile_scheme X $tile_scheme\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^large_feature_set/ ) {
$large_feature_set = 1;
print "\tLarge feature set will be used if sigs need to be trained.\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^training_images\s+(\d*)/ ) {
$training_images = $1;
print "\tNumber of training images: $training_images\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^test_images\s+(\d*)/ ) {
$test_images = $1;
print "\tNumber of test_images: $test_images\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^test_repetitions\s+(\d*)/ ) {
$test_repetitions = $1;
print "\tNumber of test repetitions: $test_repetitions\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^feature_usage_fraction\s+(\d*\.\d+)/ ) {
$feature_usage_fraction = $1;
print "\tFraction of features to be used: $feature_usage_fraction\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^html_file/ ) {
$html_file = 1;
print "\tHTML output file will be generated\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^write_fisher_weights/ ) {
$write_fisher_weights = 1;
print "\tFisher weights file will be generated\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^experiment_repetitions\s+(\d+)/ ) {
$experiment_repetitions = $1;
print "\tExperiment will be repeated $experiment_repetitions times.\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^additional_test_args\s+(.*)/ ) {
$test_args = $1;
print "\tThe following additional arguments will be applied at test time: $test_args\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^silence_wndchrm/ ) {
$silence_wndchrm = 1;
print "\tWNDCHRM will be run in silent mode.\n" if( $DEBUG1 or $verbose or $dryrun );
}
elsif( $_ =~ /^path_to_phylip\s+(\S*)/ ) {
$path_to_phylip = $1;
print "\tPath to phylip: $path_to_phylip\n" if( $DEBUG1 or $verbose or $dryrun );
}
else {
if( $DEBUG1 or $verbose or $dryrun ) {
print "\tCan't read anything from this line... Skipping.\n"
}
}
}
close CONFIG_FILE;
if( !defined $source_dir ) {
die "No source dir defined in the experiment file.";
}
if( !defined $target_dir ) {
die "No target dir defined in the experiment file.";
# print "No target dir specified. Using source directory as target dir. No experimental directory will be made.\n";
}
if( !defined $path_to_wndchrm ) {
die "Cannot continue, no path to wndchrm defined in experiment file.";
}
if( defined $tile_scheme ) {
$tile_scheme = "-t" . $tile_scheme;
} else {
$tile_scheme = "";
}
if( defined $large_feature_set ) {
$large_feature_set = "-l";
} else {
$large_feature_set = "";
}
if( defined $training_images ) {
$training_images = "-i" . $training_images;
} else {
$training_images = "";
}
if( defined $test_images ) {
$test_images = "-j" . $test_images;
} else {
$test_images = "";
}
if( defined $feature_usage_fraction ) {
$feature_usage_fraction = "-f" . $feature_usage_fraction ;
} else {
$feature_usage_fraction = "";
}
if( defined $test_repetitions ) {
$test_repetitions = "-n" . $test_repetitions;
} else {
$test_repetitions = "";
}
# The html file name and fisher weights file name are now taken care of in the loop,
# since the target directory and the experiment name change with subsequent experiment
# repetitions.
# if( defined $html_file ) {
# $html_file = "$target_dir/$experiment_name.html";
# } else {
# $html_file = "";
# }
# if( defined $write_fisher_weights ) {
# $write_fisher_weights = "-vw$target_dir/$experiment_name.fisher_weights";
# } else {
# $write_fisher_weights = "";
# }
if( ! defined $test_args ) {
$test_args = "";
}
if( defined $path_to_phylip ) {
$phylip_command = "-p" . $path_to_phylip ;
} else {
$phylip_command = "";
}
if( defined $silence_wndchrm ) {
$silence_wndchrm = " > /dev/null 2>&1" ;
} else {
$silence_wndchrm = "";
}
##################################################################################
# Run the appropriate number of experiments, given by the $experiment_repetitions parameter
##################################################################################
my $DEBUG3 = 0;
# If $experiment_repetitions = 1, then no sub directory is required
if( $DEBUG3 or $verbose or $dryrun ) {
print "Number of experiments to be run: $experiment_repetitions \n";
}
# group_row and group_col are globals that are used by the RunExperiment
# function to direct pairwise experiment traffic
my $group_row;
my $group_col;
my $original_experiment_name = $experiment_name;
my $original_target_dir = $target_dir;
my $reverse_experiment_dir;
if( $experiment_repetitions == 1 ) {
if( $pairwise_flag ){
# Make a top level directory to contain the individual experiments
$original_target_dir .= "/" . $experiment_name;
if( $DEBUG3 or $verbose or $dryrun ) {
print "Making top-level organizational directory $target_dir\n";
}
if( !$dryrun ) { mkdir "$original_target_dir" or print $! };
foreach $group_row (sort keys %group_info ) {
foreach $group_col (sort keys %group_info ) {
next if $group_row eq $group_col;
# Check to see that pairwise experiment "X vs Y" wasn't already done by
# looking to see if a "Y vs X" directory was created
$target_dir = $original_target_dir;
$reverse_experiment_dir = $target_dir . "/" . $group_col . "vs" . $group_row . "_" . $group_info{$group_col}[0] . "_VS_" . $group_info{$group_row}[0];
if( $DEBUG3 or $verbose or $dryrun ) {
print "\tChecking to see if experiment $group_row vs $group_col aready done.\n";
print "\tDoes $reverse_experiment_dir exist?\n";
}
next if -e $reverse_experiment_dir;
# The experiment name changes every time
$experiment_name = $group_row . "vs" . $group_col . "_" . $group_info{ $group_row }[0] . "_VS_" . $group_info{ $group_col }[0];
# Reset the target_dir every time you run, because $target_dir gets
# modified in the RunExperiment subroutine
if( $DEBUG3 or $verbose or $dryrun ) {
print "\tRunning pairwise experiment $group_row vs $group_col: $experiment_name\n";
}
&RunExperiment( { group_1 => $group_row, group_2 => $group_col } );
}
}
} else {
&RunExperiment;
}
}
elsif( $experiment_repetitions > 1 ) {
# Make a top level directory to contain the individual experiments
$target_dir .= "/" . $experiment_name;
if( $DEBUG3 or $verbose or $dryrun ) {
print "Making top-level organizational directory $target_dir\n";
}
if( !$dryrun ) { mkdir "$target_dir" or print $! };
for( my $ii = 1; $ii <= $experiment_repetitions; $ii++ ) {
# The experiment name changes every time
$experiment_name = $original_experiment_name . "_" . $ii;
# Reset the target_dir every time you run, because $target_dir gets
# modified in the RunExperiment subroutine
$target_dir = $original_target_dir;
if( $DEBUG3 or $verbose or $dryrun ) {
print "\tRunning experiment $ii: $experiment_name\n";
}
&RunExperiment;
}
}
##################################################################################
# Here's where all the magic happens, the linking the training and the testing
##################################################################################
my $DEBUG2 = 0;
sub RunExperiment {
if( $DEBUG2 or $verbose or $dryrun ) {
print "\n\n*******************************\nBegin experiment $experiment_name:\n";
}
my ($args) = @_;
if( $pairwise_flag ) {
if( !$args->{group_1} || !$args->{group_2} ) {
die "Internal Error: Pairwise experiment attempted, but the two groups needed weren't properly defined.\n";
} else {
print "\tPairwise analysis between group $args->{group_1} and group $args->{group_2}\n";
}
}
my $class_name;
my $dir_name;
my $num_imgs;
my @full_file_list;
my @tiff_file_list;
my $file;
my $file_stem;
my( $i, $j );
$target_dir .= "/" . $experiment_name;
if( $DEBUG2 or $verbose or $dryrun ) {
print "\tMaking experiment directory $target_dir\n";
}
if( !$dryrun ) { mkdir "$target_dir" or print $! };
ITERATE_OVER_CLASSES: foreach $class_name ( keys %class_config_info ) {
if( $DEBUG2 or $verbose or $dryrun ){
print "\tClass $class_name:\n";
}
if( $pairwise_flag ) {
# Check the %group_info hash to see if this class $class_name belongs to
# one of the groups we're checking
my $in_this_group = 0;
my @classes_in_current_groups = ( @{ $group_info{ $args->{group_1} } }, @{ $group_info{ $args->{group_2} } } );
CHECK_IF_CLASS_IN_GROUP: foreach (@classes_in_current_groups) {
if( $class_name eq $_ ) {
$in_this_group = 1;
last CHECK_IF_CLASS_IN_GROUP;
}
}
if( !$in_this_group ) {
if( $DEBUG2 or $verbose or $dryrun ){
print "\t\tSkipping over $class_name because it doesn't belong to groups $args->{group_1} and $args->{group_2}.\n";
}
next ITERATE_OVER_CLASSES;
}
}
if( $DEBUG2 or $verbose or $dryrun ) {
print "\t\tMaking directory $target_dir/$class_name\n";
}
if( !$dryrun ) { mkdir "$target_dir/$class_name" or print $! };
foreach my $slide ( @{ $class_config_info{$class_name} } ) {
$dir_name = $$slide{dir_name};
$num_imgs = $$slide{img_quant};
if( $DEBUG2 or $verbose or $dryrun ){
print "\t\t Slide \"$dir_name\", Images allocated to class $class_name: $num_imgs\n";
}
print "\t\t\tOpening $source_dir/$dir_name:\n" if( $DEBUG2 or $verbose or $dryrun );
# remember, there could be original tiffs, deconvolved tiffs with H_ or E_
# prefixes or suffixes, and also calculated sigs, all in this directory.
# This will even include the "." and ".." paths.
opendir( DIR, "$source_dir/$dir_name" ) or die "Error opening dir $dir_name";
@full_file_list = readdir(DIR);
for($i = 0; $i <= $#full_file_list; $i++ ) {
push @tiff_file_list, $full_file_list[$i] if( $full_file_list[$i] =~ /\.tif$/ ) ;
}
die "*******************************************\nFatal Error: No .tif files in directory $source_dir/$dir_name.\n" if $#tiff_file_list == -1;
warn "************************\nWarning: Only $#tiff_file_list tiffs found in $dir_name, and is less than $num_imgs specified in configuration file.\n" if( $#tiff_file_list != ($num_imgs - 1) );
# randomize the file list
for ($i = @tiff_file_list; --$i; ) {
$j = int rand ($i+1);
# print "i: $i j: $j \n";
next if $i == $j;
@tiff_file_list[$i,$j] = @tiff_file_list[$j,$i];
}
my @list_of_sigs;
# Note that the loop will bail out without error if the number of images
# asked for exceeds
for( $i = 0; $i < $num_imgs && $i <= $#tiff_file_list; $i++ ) {
# Unshift the first $num_imgs off the newly randomized list
$file = $tiff_file_list[$i];
if( $DEBUG2 or $verbose or $dryrun ) {
print "\t\t\t\tImage $i will be $file\n";
}
if( $file =~ /^(.*)\.tif$/ ) {
$file_stem = $1;
# copy the file and the sigs associated with the file
# need to be done in two operations, because image42* will copy
# not only image42.tif and image42_1.sig, but also unexpectedly
# image420, image421, etc.
if( $DEBUG2 or $verbose or $dryrun ) {
print "\t\t\t\tCreating link from $source_dir/$dir_name/$file to $target_dir/$class_name/$file\n";
}
if( !$dryrun ) {
link( "$source_dir/$dir_name/$file", "$target_dir/$class_name/$file" ) or die "Can't create hard link from $source_dir/$dir_name/$file to $target_dir/$class_name/$file: $!\n";
}
if( $DEBUG2 or $verbose or $dryrun ) {
print "\t\t\t\tCreating links for $source_dir/$dir_name/${file_stem}_*.sig to $target_dir/$class_name\n";
}
if( !$dryrun ) {
@list_of_sigs = split( "\n", `find $source_dir/$dir_name -name ${file_stem}_*.sig` );
foreach (@list_of_sigs) {
/^\S*\/(\S+)/; # This is idiomatic perl - it means perform that regular expression operation on the default variable $_, produced by foreach (@list_of_args) loop.
link( "$source_dir/$dir_name/$1", "$target_dir/$class_name/$1" ) or die "Can't create hard link from $source_dir/$dir_name/$1 to $target_dir/$class_name/$1: $!\n";
}
}
}
}
@full_file_list = ();
@tiff_file_list = ();
closedir DIR;
} # end copying images out of slides
} # end creating the classes
# moved into the loop since target_dir and experiment_name change between iterations.
if( defined $html_file && $html_file ne "" ) {
$html_file = "$target_dir/$experiment_name.html";
} else {
$html_file = "";
}
if( defined $write_fisher_weights && $write_fisher_weights ne "" ) {
$write_fisher_weights = "-vw$target_dir/$experiment_name.fisher_weights";
} else {
$write_fisher_weights = "";
}
my $cmd;
$cmd = "$path_to_wndchrm train -m $tile_scheme $large_feature_set $target_dir $target_dir/$experiment_name.fit $silence_wndchrm";
if( $DEBUG2 or $verbose or $dryrun or $silence_wndchrm) {
print "\nWndchrm train command: \n$cmd\n";
}
system( $cmd ) if( !$dryrun );
$cmd = "$path_to_wndchrm test $test_args $tile_scheme $training_images $test_images $feature_usage_fraction $test_repetitions $write_fisher_weights $phylip_command $target_dir/$experiment_name.fit $html_file $silence_wndchrm";
if( $DEBUG2 or $verbose or $dryrun or $silence_wndchrm ) {
print "\nWndchrn test command: \n$cmd\n";
}
system( $cmd ) if( !$dryrun );
# clean up by compying dend files if any to experiment dir
if( -e "$path_to_phylip/dend_file.txt" ) {
print "\tMoving $path_to_phylip/dend_file.txt to $target_dir/${experiment_name}_dend_file.txt\n" if( $DEBUG2 or $verbose or $dryrun or $silence_wndchrm);
move( "$path_to_phylip/dend_file.txt", "$target_dir/${experiment_name}_dend_file.txt" ) if !$dryrun;
} else {
print "*************************\nError: $path_to_phylip/dend_file.txt wasn't created!\n";
}
if( -e "$path_to_phylip/drawtree.infile" ) {
print "\tMoving $path_to_phylip/drawtree.infile to $target_dir/${experiment_name}_drawtree.infile\n" if( $DEBUG2 or $verbose or $dryrun or $silence_wndchrm);
move( "$path_to_phylip/drawtree.infile", "$target_dir/${experiment_name}_drawtree.infile" ) if !$dryrun;
} else {
print "*************************\nError: $path_to_phylip/drawtree.infile wasn't created!\n";
}
if( -e "$path_to_phylip/fitch.infile" ) {
print "\tMoving $path_to_phylip/fitch.infile to $target_dir/${experiment_name}_fitch.infile\n" if( $DEBUG2 or $verbose or $dryrun or $silence_wndchrm);
move( "$path_to_phylip/fitch.infile", "$target_dir/${experiment_name}_fitch.infile" ) if !$dryrun;
} else {
print "*************************\nError: $path_to_phylip/fitch.infile wasn't created!\n";
}
}
# sub SubtractIntraclassWeights
# for each class
# make a subdirectory in the target directory
# make a subsubdirectory for each slide
# copy n number of images and sigs into each subsubdirectory
# fire off wndchrm train -m and get a sub.fit file
# fireoff wndchrm test -vwWEIGHTFILE
# check that the weightfile was created, if so pack the path to an array
# for eack weightfile
# open it
#
1;
__END__
# This is a wndchrm experiment file!
# Lines that begin with a # are treated as a comment
# Experiment name - must be all one word because this becomes the name of a directory
experiment my_experiment
# Full path to your wndchrm executable
path_to_wndchrm /home/colettace/wndchrm_home_directory/wndchrm
# Source directory is the folder where your tiffs and/or sigs live
source_dir /home/colettace/where/the/images/and/sigs/live
# Target directory is where you want to create your new experiment class dir structure
# tiffs and sigs will be copied into here
target_dir /home/colettace/wndchrm/products/go/in/here
# Experiment class definition section
# directory, class to be assigned, number of images from the directory to go into class
# column 1: the word "class"
# column 2: path of class 1's images and sigs relative to source dir specified above
# column 3: the name of the class that these images belong to
# column 4: the number of images from the source directory that should be copied into the class folder
class sourcedir1 classA 90
class sourcedir2 classB 90
class sourcedir3 classC 90
class sourcedir4 classA 90
class sourcedir5 classB 90
class sourcedir6 classC 90
class an_organizing_folder/sourcedir7 classD 60
class an_organizing_folder/sourcedir8 classD 60
class an_organizing_folder/sourcedir9 classD 60
class an_organizing_folder/sourcedir10 classD 60
# The following parameters are self explanitory
tile_scheme 2 # -t4
large_feature_set # use the large feature set
training_images 89 # -i220
test_images 1 # -j20
#test_repetitions 30 # -n5
path_to_phylip /home/colettace/src/phylip-3.69
#html_file # you must explicitly ask for an html result file
feature_usage_fraction 1.0 # the fraction must be in the form 0.#, or 1.0
write_fisher_weights
experiment_repetitions 2 # Repeat the entire experiment this many times,
# generating entirely new training and test sets from the
# source image pool. Differs entirely from the -n# wndchrm switch.
# The target directory will contain a subdirectory for each
# experiment run
silence_wndchrm # run_wndchrm in silent mode
