script perl para selecionar a melhor linha genética de explosão a partir do resultado da explosão

Question

Eu tenho um arquivo enorme com saída de explosão e preciso selecionar o ID da consulta, o assunto gi e o quadro (basicamente a linha inteira) com valor e mais baixo, omitindo as linhas duplicadas (omitindo todas as outras linhas com outros valores e mais altos).Esta é a aparência do arquivo:

# BLASTX 2.2.28+
# 0 hits found
# BLASTX 2.2.28+
# Query: Tx6_c1_seq1
# Database: /mnt/swissprot
# Fields: query id, subject gi, subject title, subject length, gap opens, q. start, q. end, s. start, s. end, evalue, % subject coverage, % identity, query/sbjct frames
# 24 hits found
Tx6_c1_seq1 6439823 RecName: Full=E3 ubiquitin-protein ligase siah-1; AltName: Full=Seven in absentia homolog 1 434 1   9   173 224 282 1e-06   65  32.20   3/0
Tx6_c1_seq1 577332  RecName: Full=Putative E3 ubiquitin-protein ligase SINAT1; AltName: Full=Seven in absentia homolog 1    305 1   9   179 111 171 3e-05   67  32.79   3/0
Tx6_c1_seq1 3548505 RecName: Full=E3 ubiquitin-protein ligase siah-1; AltName: Full=Seven in absentia homolog 1 419 2   9   173 209 267 8e-05   65  32.20   3/0
Tx6_c1_seq1 577547  RecName: Full=E3 ubiquitin-protein ligase siah2; AltName: Full=Seven in absentia homolog 2; AltName: Full=Xsiah-2   313 1   15  173 125 181 2e-04   62  29.82   3/0
Tx6_c1_seq1 577417  RecName: Full=E3 ubiquitin-protein ligase Siah1; AltName: Full=Seven in absentia homolog 1; Short=Siah-1    282 1   15  173 96  152 3e-04   62  29.82   3/0
Tx6_c1_seq1 577554  RecName: Full=E3 ubiquitin-protein ligase SINAT2; AltName: Full=Seven in absentia homolog 2 308 1   9   179 114 174 4e-04   67  31.15   3/0
# BLASTX 2.2.28+
# Query: Tx_11_c0_seq1
# Database: /mnt/swissprot
# Fields: query id, subject gi, subject title, subject length, gap opens, q. start, q. end, s. start, s. end, evalue, % subject coverage, % identity, query/sbjct frames
# 1 hits found
Tx_11_c0_seq1   977285  RecName: Full=120.7 kDa protein in NOF-FB transposable element  1056    15  957 28  147 455 8e-13   79  27.81   -2/0
# BLASTX 2.2.28+
# Query: Tx_11_c1_seq1

A saída esperada neste caso deve ser apenas estas duas linhas, pois são as que têm menor e_value:

Tx6_c1_seq1 6439823 RecName: Full=E3 ubiquitin-protein ligase siah-1; AltName: Full=Seven in absentia homolog 1 434 1   9   173 224 282 1e-06   65  32.20   3/0
Tx_11_c0_seq1   977285  RecName: Full=120.7 kDa protein in NOF-FB transposable element  1056

Tenho meu código escrito, mas parece que não funciona.Vocês poderiam me ajudar a resolver esse problema.Eu realmente apreciaria seu tempo e ajuda.Isto é o que tenho até agora:

#!/usr/bin/perl -w

# Author:
# 01/07/2014
# This script removes duplicate records from a "short" format BLAST output file, and keeps only the "best" records  (sorts by smallest e-value and then biggest percent identity)
# Usage: bestblast.pl <input file> <output file>

#-----------------------------------------------------------------------------------------------------------------------------------------
#Deal with passed parameters
#-----------------------------------------------------------------------------------------------------------------------------------------
#If no arguments are passed, show usage message and exit program.
if ($#ARGV == -1) {
    usage("BLAST BEST 1.0 2014");
    exit;
}

#get the names of the input file (first argument passed) and output file (second argument passed)
$in_file = $ARGV[0];
$out_file = $ARGV[1];

#Open the input file for reading, open the output file for writing.
#If either are unsuccessful, print an error message and exit program.
unless ( open(IN, "$in_file") ) {
    usage("Got a bad input file: $in_file");
    exit;
}
unless ( open(OUT, ">$out_file") ) {
    usage("Got a bad output file: $out_file");
    exit;
}

#Everything looks good. Print the parameters we've found.
print "Parameters:\ninput file = $in_file\noutput file = $out_file\n\n";

#-----------------------------------------------------------------------------------------------------------------------------------------
#The main event
#-----------------------------------------------------------------------------------------------------------------------------------------

$counter = 0;
$total_counter = 0;

print "De-duplicating File...\n";

@in = <IN>;

#Do stuff for each line of text in the input file.
foreach $line (@in) {
    #if the line starts with a pound symbol, it is not real data, so skip this line.
    if ( $line =~ /^#/ ) {
     next;
     }

    #Count the total number of data lines in the file.
    $total_counter++;

    #The chomp commands removes any new line (and carriage return) characters from the end of the line.
    chomp($line);

    #Split up the tab delimited line, naming only the variables we are interested in (i.e. query id, subject gi, subject title, subject length, gap opens, q. start, q. end, s. start, s. end, evalue, % subject coverage, % identity, query/sbjct frames)
    ($query_id, $subject_gi, $subject_title, $subject_length, $gap_opens, $q_start, $q_end, $s_start, $s_end, $evalue, $subject_coverage, $identity, $query_sbjct_frames) = split(/\t/, $line);

    #check to see if the id label is already in the list of ids (called dedupe)
    #if its not there, add it.
    if ( $dedupe{$query_id} ) {
    #if it is, look at the old line to see if it is still "better" than the new one.
    ($query_id, $subject_gi, $subject_title, $subject_length, $gap_opens, $q_start, $q_end, $s_start, $s_end, $list_evalue, $subject_coverage, $list_identity, $query_sbjct_frames) = split(/\t/,$dedupe{$query_id});

    #if the new evalue is better than the old one, change the value of this id to the new line.
    #otherwise, if the the new evalue is the same, and the percent_identity is better, change the value of this id to the new line.
    #otherwise, don't do anything (keep the old line).
    if ( $evalue < $list_evalue ) {
        $dedupe{$query_id} = $line;
    }
    elsif ( $evalue == $list_evalue ) {
        if ( $identity > $list_identity ) {
        $dedupe{$query_id} = $line;
        }
    }
    }
    else {
    $dedupe{$query_id} = $line;
    #count the number of non-duplicated lines we have.
    $counter++;
    }
}
print "Total # records = $total_counter\nBest only # records = $counter\n";
print "Writing to output file...\n";

#Print the final "dedupe" list to the new file (adding the new line back on the end).
foreach $query_id (sort keys %dedupe) {
    print OUT "$dedupe{$query_id}\n";
}

#Close the files.
close(IN);
close(OUT);
print "Done.\n";

#-----------------------------------------------------------------------------------------------------------------------------------------
#Subroutines
#-----------------------------------------------------------------------------------------------------------------------------------------
sub usage {
    my($message) = @_;
    print "\n$message\n";

    print "\nThis script removes duplicate records from a \"short\" format BLAST output file, and keeps only the \"best\" records.\nIt sorts by smallest e-value and then biggest percent identity.\n";
    print "Usage: bestbenter code herelast.pl <input file> <output file>\n";
    print "\n Author \n";
    print "01/07/2014\n";
}

Answer 1

Tente adicionar "use strict" na parte superior após o shebang - isso pode ajudar a localizar outra coisa.

Tente substituir "if($dedupe{$query_id})" por "if(definido($dedupe{$query_id}))"

Tenha em mente que a maioria das pessoas no SO não são biólogos/genomistas (!) e não têm ideia do que você está falando, apenas vemos números e palavras que não significam nada para nós, então se você puder explicar melhor, talvez possamos ajude mais.

A propósito, o seguinte é mais idiomático.

next if $line =~ /^#/;

Seu código sempre vai da linha 64 à 81.Ele nunca entra nos testes secundários - nunca encontra duplicatas.Tente executá-lo no depurador, com isto:

perl -d yourprog INFILE OUTFILE

em seguida, faça "n" para "próxima linha" repetidamente.Você pode imprimir valores de variáveis ​​com "p nome da variável".

Se eu alterar o separador dos split()s de tabulação para espaço, recebo isto - que tem pelo menos o número correto de registros de saída:

De-duplicating File...
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "AltName:" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "in" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "Full=Seven" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "absentia" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "Full=Seven" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "absentia" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "in" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "Full=Seven" isn't numeric in numeric lt (<) at ./go line 71, <IN> line 21.
Argument "homolog" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Argument "absentia" isn't numeric in numeric gt (>) at ./go line 75, <IN> line 21.
Total # records = 7
Best only # records = 2
Writing to output file...
Done.
iMac:~/tmp: more out
Tx6_c1_seq1 6439823 RecName: Full=E3 ubiquitin-protein ligase siah-1; AltName: Full=Seven in absentia homolog 1 434 1   9   173 224 282 1e-06   65  32.20   3/0
Tx_11_c0_seq1   977285  RecName: Full=120.7 kDa protein in NOF-FB transposable element  1056    15  957 28  147 455 8e-13   79  27.81   -2/0

Answer 2

Por que você está tentando escrever seu próprio analisador de explosão?Usar BioPerl

http://www.bioperl.org/wiki/HOWTO:SearchIO#NCBI-BLAST_parsing_problems

Eu não uso mais Perl, mas aqui está uma ideia aproximada do que fazer

while (my $result = $report->next_result) {
    print "Query: ".$result->query_name."\n";
    while (my $hit = $result->next_hit) {
        while ($hsp = $hit->next_hsp) {
            my evalue = $hsp->evalue;
            #convert to decimal notation
            $decimal_notation = sprintf("%.10g", $scientific_notation);

            ##... i'll leave the rest up to you
        }
     }
}

Answer 3

evalue está em notação científica, perl irá tratá-lo como uma string ao fazer a comparação menor que.

Acho que também faria as coisas de desduplicação de forma diferente ...