#!/usr/bin/perl use strict; #use warnings; use lib qw(/local/Projects/Perl/casp13/src/scripts/evaluation_scripts/Parsers); use LocalConfig; use EvaluationResult; use ResultsParser; use Getopt::Long; use lib qw(Classes); use TargetsManager; ########################################################### # # The script parses the corresponding directories with # results and creates file containing table of scores. # (One file per target). These files are used to perform # analysis in R. # ########################################################### # VARS my $DEBUG = 0; my $headerDone = 0; # flag to print header just once my $TARGET; my $OUTFILE; # output file in csv format with my $TMP_FILE = "/tmp/tmp_".$$.".csv"; # tmp_file for parsing results from outputs; it is used as input to Rscript my @models; # array of models my @preds; # array of members of class Evaluator() - one evaluator for eveyr model my $Rscript = "/local/Projects/Perl/casp13/src/scripts/evaluation_scripts/calcZscores.R"; # read arguments GetOptions ( "target|t=s" => \$TARGET, "output|o=s" => \$OUTFILE ); # check args if (!defined($TARGET)){ &usage(); } if (!defined($OUTFILE)){ $OUTFILE = sprintf("/local/CASP13/tmp_Grishin/%s.GrishinScores", $TARGET); } my %ALGROUPS = ( '006'=>1, '008'=>1, '011'=>1, '014'=>1, '022'=>1, '023'=>1, '024'=>1, '026'=>1, '032'=>1, '034'=>1, '038'=>1, '040'=>1, '041'=>1, '042'=>1, '044'=>1, '049'=>1, '050'=>1, '054'=>1, '056'=>1, '063'=>1, '064'=>1, '065'=>1, '067'=>1, '073'=>1, '076'=>1, '080'=>1, '092'=>1, '097'=>1, '106'=>1, '110'=>1, '111'=>1, '116'=>1, '117'=>1, '118'=>1, '120'=>1, '128'=>1, '132'=>1, '133'=>1, '144'=>1, '145'=>1, '153'=>1, '155'=>1, '156'=>1, '157'=>1, '160'=>1, '162'=>1, '169'=>1, '171'=>1, '173'=>1, '184'=>1, '186'=>1, '192'=>1, '193'=>1, '197'=>1, '203'=>1, '204'=>1, '206'=>1, '210'=>1, '212'=>1, '216'=>1, '228'=>1, '230'=>1, '235'=>1, '237'=>1, '241'=>1, '251'=>1, '258'=>1, '260'=>1, '263'=>1, '268'=>1, '269'=>1, '276'=>1, '277'=>1, '279'=>1, '281'=>1, '282'=>1, '290'=>1, '296'=>1, '300'=>1, '301'=>1, '310'=>1, '311'=>1, '317'=>1, '322'=>1, '326'=>1, '328'=>1, '333'=>1, '335'=>1, '336'=>1, '338'=>1, '340'=>1, '342'=>1, '345'=>1, '346'=>1, '347'=>1, '349'=>1, '357'=>1, '358'=>1, '360'=>1, '361'=>1, '362'=>1, '364'=>1, '368'=>1, '381'=>1, '385'=>1, '386'=>1, '391'=>1, '401'=>1, '403'=>1, '404'=>1, '410'=>1, '414'=>1, '417'=>1, '419'=>1, '420'=>1, '425'=>1, '428'=>1, '430'=>1, '433'=>1, '434'=>1, '436'=>1, '437'=>1, '438'=>1, '439'=>1, '442'=>1, '445'=>1, '448'=>1, '452'=>1, '454'=>1, '457'=>1, '460'=>1, '465'=>1, '466'=>1, '476'=>1, '479'=>1, '482'=>1, '483'=>1, '490'=>1, '492'=>1, '493'=>1, '495'=>1, '499'=>1, '077'=>1, '486'=>1 ); my %SERVERS = ( '008'=>1, '011'=>1, '022'=>1, '038'=>1, '041'=>1, '050'=>1, '073'=>1, '110'=>1, '117'=>1, '133'=>1, '145'=>1, '156'=>1, '160'=>1, '171'=>1, '184'=>1, '193'=>1, '206'=>1, '210'=>1, '212'=>1, '216'=>1, '228'=>1, '237'=>1, '251'=>1, '263'=>1, '268'=>1, '277'=>1, '279'=>1, '300'=>1, '335'=>1, '345'=>1, '346'=>1, '349'=>1, '381'=>1, '410'=>1, '414'=>1, '420'=>1, '436'=>1, '448'=>1, '452'=>1, '454'=>1, '466'=>1, '479'=>1, '492'=>1, '499'=>1 ) ; my %TCGROUPS = ( '032'=>1, '038'=>1, '040'=>1, '041'=>1, '042'=>1, '044'=>1, '064'=>1, '065'=>1, '080'=>1, '155'=>1, '157'=>1, '162'=>1, '169'=>1, '186'=>1, '219'=>1, '276'=>1, '287'=>1, '300'=>1, '310'=>1, '329'=>1, '342'=>1, '345'=>1, '357'=>1, '361'=>1, '420'=>1, '428'=>1, '476'=>1, '479'=>1, '490'=>1 ); my %GROUPS; my %DONE_GROUPS; my $targetManager = new TargetsManager(); my $target_name; my $domain; if ($TARGET =~ m/^(T[acspx0-9][0-9]{3})-D([1-9]$)/){ $target_name = $1; $domain = $2; } else { $target_name = $TARGET; } my $targ_id = $targetManager->get_id_by_name($target_name); my %info = $targetManager->info($targ_id); if ($target_name =~ m/^T[acspx]/){ %GROUPS = %TCGROUPS; } elsif ($info{IS_SERVER_ONLY} ){ %GROUPS = %SERVERS; } else { %GROUPS = %ALGROUPS; } # read list of models my @models = &getModels(); if ($DEBUG){ foreach my $m (@models){ print $m."\n"; } } # end read predictions my $no_scores; open TMP, ">$TMP_FILE" || die("Can not open output file for writing\n"); # parse scores of all predictions and write to output file my @preds; # array of all predictions foreach my $m (@models){ my $parser = new ResultsParser(); my $evaluator; # parse QCS results file $evaluator = $parser->parseQCS($m); # parse SOV result file $evaluator = $parser->parseSOV($evaluator); # parse CE result file : obsolete score # $evaluator = $parser->parseCE($evaluator); # parse TMscore $evaluator = $parser->parseTMscore($evaluator); # parse Mammoth result file $evaluator = $parser->parseMammoth($evaluator); # parse LGA_SDA result file $evaluator = $parser->parseLGA_SDA($evaluator); # parse LGA_SIA result file $evaluator = $parser->parseLGA_SIA($evaluator); # parse DaliLite result file $evaluator = $parser->parseDali($evaluator); # parse grishin conts result file $evaluator = $parser->parseGrContS($evaluator); if ($DEBUG){ # printf("%s,%s%s,%s\n", $evaluator->createTargetName(), $evaluator->{_pfrmat}, $evaluator->{_gr_code}, $evaluator->{_model}); # print "contS\t".$evaluator->{_scores}{contSintra}."\n"; # print "handedness\t".$evaluator->{_scores}{handedness}."\n"; # print "qcs\t".$evaluator->{_scores}{qcs}."\n"; # print "sov\t".$evaluator->{_scores}{sov}."\n"; # print "tm\t".$evaluator->{_scores}{tm}."\n"; # print "tm_n_align\t".$evaluator->{_scores}{tm_n_align}."\n"; # print "mammoth_ln_e\t".$evaluator->{_scores}{mammoth_ln_e}."\n"; # print "mammoth_n_align\t".$evaluator->{_scores}{mammoth_n_align}."\n"; # print "lga_gdt_ts\t".$evaluator->{_scores}{lga_gdt_ts}."\n"; # print "lga_n_align\t".$evaluator->{_scores}{lga_n_align}."\n"; # print "dali\t".$evaluator->{_scores}{dali}."\n"; # print "dali_n_align\t".$evaluator->{_scores}{dali_n_align}."\n"; # last; } my $header = "Target,GrCode,Model"; my $record = sprintf("%s,%s%s,%s%s", $evaluator->createTargetName(), $evaluator->{_pfrmat}, $evaluator->{_gr_code}, $evaluator->{_model}, (defined($evaluator->{_parent}) ? '_'.$evaluator->{_parent}: '')); $no_scores = scalar(keys %{$evaluator->{_scores}}); foreach my $score (sort keys %{$evaluator->{_scores}}){ $header .= sprintf(",$score"); $record .= sprintf(",%s", defined($evaluator->{_scores}{"$score"}) ? $evaluator->{_scores}{"$score"} : 'NA'); } $header .= ",missing\n"; $record .= ",0\n"; print TMP $header if ($headerDone == 0); $headerDone = 1; print TMP $record; $DONE_GROUPS{sprintf("%03d", $evaluator->{_gr_code})} = 1; } # end read predictions and write output # write missing groups foreach my $key (sort keys %GROUPS){ if (defined($DONE_GROUPS{$key})){next;} printf TMP "%s,TS%03d,1", $TARGET, $key; for(my $c=0; $c < $no_scores; $c++){ print TMP ",NA"; } print TMP ",NA\n"; } close TMP; if ($DEBUG){exit;} # run Rscript and calculate z scripts # for first models my $program = sprintf("%s %s %s first", $Rscript, $TMP_FILE, $OUTFILE); system("$program"); # for best models $program = sprintf("%s %s %s best", $Rscript, $TMP_FILE, $OUTFILE); system("$program"); # for all models #$program = sprintf("%s %s %s all", $Rscript, $TMP_FILE, $OUTFILE); #system("$program"); # clean if (-e $TMP_FILE){ system("rm $TMP_FILE"); } # change groupand permissions system("chgrp casp $OUTFILE* ; chmod g+w $OUTFILE*"); exit; #---------------------- # SUBROUTINES #---------------------- # the subroutine returns list of models for which the results # should be retrieved # it parses the directory: /QCS// sub getModels{ my @result; my $dir = sprintf("%s/QCS/%s", $LOCAL_CONFIG->{RESULTS_DIR}, $TARGET); opendir D, $dir; while(defined(my $f = readdir(D))){ next unless $f =~ m/\.qcs$/; $f =~ s/\.qcs//; push @result, $f; } closedir(D); @result = sort @result; return @result; } sub usage{ print " ====== USAGE: $0 --target \n \t - required argument - target name; The script parses result files in directory /local/CASP13/RESULTS/; and creates file .GrishinScores.csv. The output file is use to perform atatistical analysis in R. The list of programs for which the script expects the results to exist in the corresponding directories: lga_sda - /LGA/SDA/DATA//, mammoth - /MAMMOTH//, daliLite - /DaliLite// ce - /CE//, sov - /SOV//, qcs - /QCS//. The list of all models is parsed from /QCS//.\n =========\n"; exit; } 1;