view release on metacpan or  search on metacpan

bin/geoCancerPrognosticDatasetsRetriever  view on Meta::CPAN

######################################################################\n\n\n" , color("reset");
}
############################ SUBROUTINE 4 #######################################################
#This subroutine checks all command line input switches and arguments and warns user if something 
#is missing.
sub input_parameters_check {

my $error_message = "Error: The following arguments are missing: CANCER_TYPE PLATFORM_CODES\n";
my $help_message1  = "Usage: geoCancerPrognosticDatasetsRetriever -d \"CANCER_TYPE\" -p \"PLATFORMS_CODES\"";
my $help_message2  = "Mandatory arguments:
  CANCER_TYPE           type of the cancer as query search term
  PLATFORM_CODES        list of GPL platform codes

  Optional arguments:
  -h                    show help message and exit";
 
	#parse command line switches and their arguments into a hash.
	getopts("hd:p:", \%options);
	
    #Check for help switch and, if present, output help text.
	if ($options{h}) {

		print color ("green"), "$help_message1", color("reset");
		print color ("green"), "\n$help_message2\n", color("reset");
		exit;
	}
	
	elsif ($options{d} and $options{p}) {
		
		print color ("green"), "Checking input parameters...", color("reset");
		mini();
	}
	
	elsif (!$options{d} or !$options{p}) {
			
		print color ("green"), "$help_message1\n$help_message2\n", color("reset");
		print color ("red"), $error_message, color("reset");
		exit;	
	} 
	
	sub mini {
		
		print color ("green"), "done\n", color("reset");
		
		my $restart_input_file;
		my $temp_filename = "$options{d}";
		my ($query_term_1, $query_term_2) = split ( / /, $temp_filename );
		#add dash in cancer type query search term.
		$cancer_type = uc ( join ( '-', $query_term_1, $query_term_2 ) );
		my $cancer = "$query_term_1";
		my @files = glob("$prog_path/data/$cancer\_cancer_GEO_*.txt");
		my @sorted_files = sort {$b cmp $a} @files;
		$run_dir = "$prog_path/results/$cancer_type\_GEO-files";
		
		foreach my $file (@sorted_files) {
			
			$restart_input_file = basename($file);
			last;
		}

		#If an old run file was found, prompt the user with choices to make.
		if (-e "$run_dir") {
			
			print color ("red"), "$cancer_type\_GEO-files directory exists...This run was not completed\n", color("reset");
			my $text = "";
			my $ok = timed_response( sub { 	
			
				print color ("red"), "Do you want to resume an interrupted execution [r], or start a new one [n]? (r/n)\nDefault selection will be [n] after 10 seconds...\n", color("reset"); $text = <STDIN>; 
			
			}, 10);
			
			chomp($text);

			if ($text eq "r") {
				
				print color ("green"), "Resuming analysis using input file: $restart_input_file\n", color("reset");
				$platform_gpl= uc($options{p});
				my $regex1 = join( '', ( split(/GPL/, $platform_gpl) ) );
				$regex_platform = join( '|', ( split(/ /, $regex1) ) );
				$input_file = $restart_input_file;
				$output_file = "$cancer_type.out";
			}
			
			#this is when the user selects "n", or types nothing/ or 10 seconds elapse -> defaults to "n"
			else {

				print color ("green"), "Starting new analysis...\n", color("reset");
				system ("rm -r $run_dir"); #remove old results output directory
				new_run($query_term_1);
			}

			sub timed_response {

				my ($f, $sec) = @_;

				return eval {
			  
					local $SIG{ALRM} = sub { die };
					alarm($sec);
					$f->();
					alarm(0);
					1;
				};
			}	
		}

		#else no "interrupted" run directory was found. Start a new run.
		else {
			
			new_run($query_term_1);
		}

		sub new_run {
			
			my $cancer = $_[0];
			print color ("green"), "Downloading input file for \"$cancer\" cancer from GeoDatasets...", color("reset");
			$input_file  = download_geo_input($options{d});
			print color ("green"), "done\n", color("reset");
			system ("mkdir $run_dir"); #create results output directory
			$platform_gpl= uc($options{p});
			my $regex1 = join( '', ( split(/GPL/, $platform_gpl) ) );

bin/geoCancerPrognosticDatasetsRetriever  view on Meta::CPAN

    
    #Check for a GeoDatasets timeout error and abort run, if found.
    if (!$data) {
		
		print color ("red"), "\nThe download from GeoDatasets was not successful...\nA GeoDatasets timeout error was detected: current run aborted...\nPlease restart the run...\n", color("reset");
		exit; #abort current run
	}

    #add date & time to current input file download
	my $geo_datasets_file = "$cancer\_cancer_GEO_$current_date_time.txt"; 
    
    open(FH, ">$prog_path/data/$geo_datasets_file") or die "Cannot open file for writing the GDS input:$!\n";
    binmode(FH, ":utf8");
    
    print FH "$data";

    close(FH);
    
    return $geo_datasets_file;
}
############################ SUBROUTINE 6 #######################################################
#This subroutine performs minor formatting of a GEO input file to merge the title and abstract 
#lines together to prevent the regex lines from missing potential keyword hits in the 'title' 
#line.
sub format_input {
	
	my $raw_input 	=	$_[0];
	my $out_file  	=	$_[1];
	my $concatenate;
	
	print color ("green"), "Formatting Input: $input_file...", color("reset");
	
	open (IN, "$prog_path/data/$raw_input") or die "Cannot open file for reformatting: $raw_input. $!.\n";
	open (OUT, '>', "$prog_path/data/$out_file") or die "Cannot open file for writing reformatted data: $out_file $!\n";
	
	while ($line = <IN>) {
			
		#title line check only
		if ($line =~ m/(^\d+\.\s+.*)/) { 
			
			$concatenate = $line;
			chomp($concatenate);
		}
 			
		#abstract line
		elsif ($line !~ m/(^\d+\.\s+.*)/) {
			
			$concatenate .= "$line";
			print OUT "$concatenate";
			$concatenate  = ""; #reinitialize variable for next entry.
		}	
	}
	
	print color ("green"), "done\n", color("reset");
	
	close (IN);
	close (OUT);
}
############################ SUBROUTINE 7 #######################################################
#Check for the presence of curl in the $PATH. If not found, install on an Ubuntu system or if 
#not Ubuntu, prompt user to install it manually.
sub check_curl {
	
	#check for the presence of curl binary
	my $check = qx{which curl};
	
	#if no curl binary was found, install it on Ubuntu/Ubuntu-based systems
	if (!$check) {
			
		#check if current system is Ubuntu/or Ubuntu-based
		my $ubuntu = qx{uname -a};
			
		if ($ubuntu=~ /.+ubuntu.+/ig) {
			
			print color ("red"), "curl binary was not found: follow onscreen instructions/input your password for its installation...\n\n", color("reset");
			system("sudo apt -y install curl"); #install curl
			print "done\n";	
		} 
			
		else { 
				
			print "curl is not found on this system: install it on your system.\n"; 
		}	
	}
}
############################ SUBROUTINE 8 #######################################################
#This subroutine runs the main processing steps, while running other subroutines to continue the 
#processing pipeline.
sub main {
	
	my $main_formatted_input_file  = $_[0];
	my $main_output_file           = $_[1];
	
	print color ("green"), "Analyzing Input: $main_formatted_input_file file...\n", color("reset");

	#open input file
	open (FH, "$prog_path/data/$main_formatted_input_file") or die "Cannot open file: $main_formatted_input_file $!\n";

	#open output file
	open (FH2, '>', "$prog_path/results/$main_output_file") or die "Cannot open file for writing data: $!\n";
    
	while ($line = <FH>) {
			
		if ($line =~ m/.*(prognosis|prognostic|prognostically|prognosticator|survival|survive|survives|survived|surviving).*/ig) { 
		#if ($line =~ m/.*(progno.+\s?|surviv.+\s?).*/ig) {
			
			$flag = 1; 
			$prog_flag = 1; 
			next; 
		}
		#this conditional activates when the above keywords are not found and only "more..." is found. 
		#Note this conditional implicitly doesn't get executed if both the desired keyword and "more..." are found.
		elsif ($line =~ m/.+(more\.\.\.)/ig) { 	
			
			$flag = 1; 
			$wget_flag = 1; 
			next;
		}
		
		elsif ($line =~ m/(^Organism:\s+Homo\s+sapiens.*)/ig) {