# This package is for handling ids files, and the
# files distributed below /projects/compbio/experiments/models.97/pdb/
package ChainIds;
use English;
use FileHandle;
use Exporter;
@ISA = qw(Exporter);
@EXPORT = qw(	ReadIDs
		ReadIDsFH
                PrintIDsStdout
		dir_path_for_id
		pretty_dir_path_for_id
		subdir_for_alignspec
		get_suffix
	        full_existing_filename
		pdb_file
	);

use vars qw(	$models97
		$indexesdir
		$models97_subdir
		$chain_subdir
		%SUBDIR
		%override_suffix
	);
    
    $models97 = $ENV{MODELS97} ? $ENV{MODELS97} :
		"/projects/compbio/experiments/models.97";
    $indexesdir = $ENV{INDEXES97} ? $ENV{INDEXES97} :
		"/projects/compbio/experiments/models.97/indexes";
    
    #########################################################################
    # Here is the section that specifies the location of .a2m files.
    # The keys to %SUBDIR are the legal values for -a2m.
    # Even if the alignment files are gzipped, specify $a2m_suffix
    # without the ".gz".
    #########################################################################
    $SUBDIR{"fnok"} = "struct-align";    
    $SUBDIR{"fssp"} = "struct-align";    
    $SUBDIR{"fssp-t98"} = "struct-align";    
    $SUBDIR{"hssp"} = "struct-align";    
    $SUBDIR{"fssp-thin20.pdbnums.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-chi1-3.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-abs_sa-CB-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-length-4.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-near-backbone-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-near-backbone-11.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-way-back-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-way-back-11.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-16-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-14-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-10-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-8-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CB-burial-6.5-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CA-burial-14-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CA-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CA-burial-10-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CA-burial-8-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-CA-burial-6.5-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-wet-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-wet-burial-10-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-wet-burial-8-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-wet-burial-6.5-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-dry-burial-14-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-dry-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-dry-burial-10-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-dry-burial-8-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-dry-burial-6.5-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-12-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-10-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-8-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-9-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-6.5-7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-burial-6.5-9.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-rel_sa10.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-rel_sa7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-rel_sa3.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-rel_sa2.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20-abs_sa7.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.tco.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.str.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.str2.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.akp.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.alpha.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.pmt.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.seq.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.2d.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.stride.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.ang.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.co.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.pb.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.gamma.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin20.kappa.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin30.tco.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin30.seq.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin30.2d.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin30.stride.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin40.tco.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin40.seq.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin40.2d.rdb"} = "struct-align";
    $SUBDIR{"fssp-thin40.stride.rdb"} = "struct-align";


    $SUBDIR{"target96"} = "nostruct-align";    
    $SUBDIR{"target97"} = "nostruct-align";    
    $SUBDIR{"target98"} = "nostruct-align";    
    $SUBDIR{"target98_1"} = "nostruct-align";    
    $SUBDIR{"target98_2"} = "nostruct-align";    
    $SUBDIR{"target98_3"} = "nostruct-align";    
    $SUBDIR{"target98-pdb"} = "nostruct-align";
    $SUBDIR{"target98-mixed"} = "nostruct-align";
    $SUBDIR{"target98-dssp"} = "nostruct-align";
    $SUBDIR{"t99"} = "nostruct-align";
    $SUBDIR{"t99-thin90"} = "nostruct-align";
    $SUBDIR{"t2k"} = "nostruct-align";
    $SUBDIR{"t2k-thin90"} = "nostruct-align";
    $SUBDIR{"t2k-thin62"} = "nostruct-align";
    $SUBDIR{"t04"}        = "nostruct-align";
    $SUBDIR{"t04-thin90"} = "nostruct-align";
    $SUBDIR{"t04-thin62"} = "nostruct-align";
    
    $SUBDIR{"t06"}        = "nostruct-align";
    $SUBDIR{"t06-thin90"} = "nostruct-align";
    $SUBDIR{"t06-thin62"} = "nostruct-align";
    
    $SUBDIR{"t99-w0.5.mod"} = "nostruct-align";
    $SUBDIR{"t2k-w0.5.mod"} = "nostruct-align";
    $SUBDIR{"t04-w0.5.mod"} = "nostruct-align";
    $SUBDIR{"t06-w0.5.mod"} = "nostruct-align";

    $SUBDIR{"t99-ebghtl.mod"} = "nostruct-align";
    $SUBDIR{"t2k-ebghtl.mod"} = "nostruct-align";
    $SUBDIR{"t04-ebghtl.mod"} = "nostruct-align";

    $SUBDIR{"t99-str2.mod"} = "nostruct-align";
    $SUBDIR{"t2k-str2.mod"} = "nostruct-align";
    $SUBDIR{"t04-str2.mod"} = "nostruct-align";

    $SUBDIR{"t99-pb.mod"} = "nostruct-align";
    $SUBDIR{"t2k-pb.mod"} = "nostruct-align";
    $SUBDIR{"t04-pb.mod"} = "nostruct-align";

    $SUBDIR{"t99-2d.mod"} = "nostruct-align";
    $SUBDIR{"t2k-2d.mod"} = "nostruct-align";
    $SUBDIR{"t04-2d.mod"} = "nostruct-align";

    $SUBDIR{"t2k-ebghtl-logo.eps"} = "nostruct-align";
    $SUBDIR{"t2k-test-100-30-rel_sa7-db.mlib"} = "nostruct-align";
    
    $SUBDIR{"hssp"} = "struct-align";
    $SUBDIR{"dssp"} = "info";    
    $SUBDIR{"stride"} = "info";    
    $SUBDIR{"dssp-mixed"} = "info";    
    $SUBDIR{"stride-mixed"} = "info";    
    $SUBDIR{"guide"} = "info";    
    $SUBDIR{"domainseq"} = "info";    


    $SUBDIR{"pdb.seq"} ="info";
    $SUBDIR{"dssp.seq"} ="info";
    $SUBDIR{"stride.seq"} ="info";
    $SUBDIR{"dssp-mixed.seq"} ="info";
    $SUBDIR{"stride-mixed.seq"} ="info";
    $SUBDIR{"hssp.guide"} ="info";

    $SUBDIR{"dssp.asa"} ="info";
    $SUBDIR{"dssp.pmt"} ="info";
    $SUBDIR{"dssp.int"} ="info";
    $SUBDIR{"pdb.2d"} ="info";
    $SUBDIR{"dssp.2d"} ="info";
    $SUBDIR{"stride.2d"} ="info";
    $SUBDIR{"dssp-mixed.2d"} ="info";
    $SUBDIR{"stride-mixed.2d"} ="info";
    $SUBDIR{"stride-mixed.dssp"} ="info";
    $SUBDIR{"hssp.2d"} ="info";

    $SUBDIR{"dssp.ang"} ="info";
    $SUBDIR{"stride-mixed.ang"} ="info";
    $SUBDIR{"stride-mixed.bys"} ="info";
    
    $SUBDIR{"dssp.co"} ="info";
    $SUBDIR{"stride-mixed.co"} ="info";

    $SUBDIR{"dssp.rot"} ="info";
    $SUBDIR{"stride-mixed.rot"} ="info";
    
    $SUBDIR{"dssp.ppi"} ="info";
    $SUBDIR{"stride-mixed.ppi"} ="info";
    
    $SUBDIR{"dssp.tco"} = "info";
    $SUBDIR{"stride-mixed.tco"} = "info";

    $SUBDIR{"dssp.alpha"} = "info";
    $SUBDIR{"stride-mixed.alpha"} = "info";

    $SUBDIR{"dssp.pb"} = "info";
    $SUBDIR{"stride-mixed.pb"} = "info";

    $SUBDIR{"dssp.gamma"} = "info";
    $SUBDIR{"stride-mixed.gamma"} = "info";

    $SUBDIR{"dssp.kappa"} = "info";
    $SUBDIR{"stride-mixed.kappa"} = "info";

    $SUBDIR{"dssp.str"} = "info";
    $SUBDIR{"stride-mixed.str"} = "info";

    $SUBDIR{"dssp.str2"} = "info";
    $SUBDIR{"stride-mixed.str2"} = "info";

    $SUBDIR{"dssp.akp"} = "info";
    $SUBDIR{"stride-mixed.akp"} = "info";

    $SUBDIR{"dssp.ehl2-stride"} = "info";
    $SUBDIR{"stride-mixed.ehl2-stride"} = "info";

    $SUBDIR{"dssp.ehl2-dssp"} = "info";
    $SUBDIR{"stride-mixed.ehl2-dssp"} = "info";

    $SUBDIR{"stride-mixed.chi1-3"} = "info";

    $SUBDIR{"stride-mixed.abs_sa-CB-burial-12-7"} = "info";

    $SUBDIR{"stride-mixed.length-4"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-16-7"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-14-7"} = "info";
    $SUBDIR{"stride-mixed.near-backbone-7"} = "info";
    $SUBDIR{"stride-mixed.near-backbone-11"} = "info";
    $SUBDIR{"stride-mixed.way-back-7"} = "info";
    $SUBDIR{"stride-mixed.way-back-11"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-12-7"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-10-7"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-8-7"} = "info";
    $SUBDIR{"stride-mixed.CB-burial-6.5-7"} = "info";
    $SUBDIR{"stride-mixed.CA-burial-14-7"} = "info";
    $SUBDIR{"stride-mixed.CA-burial-12-7"} = "info";
    $SUBDIR{"stride-mixed.CA-burial-10-7"} = "info";
    $SUBDIR{"stride-mixed.CA-burial-8-7"} = "info";
    $SUBDIR{"stride-mixed.CA-burial-6.5-7"} = "info";
    $SUBDIR{"stride-mixed.wet-burial-12-7"} = "info";
    $SUBDIR{"stride-mixed.wet-burial-10-7"} = "info";
    $SUBDIR{"stride-mixed.wet-burial-8-7"} = "info";
    $SUBDIR{"stride-mixed.wet-burial-6.5-7"} = "info";
    $SUBDIR{"stride-mixed.dry-burial-14-7"} = "info";
    $SUBDIR{"stride-mixed.dry-burial-12-7"} = "info";
    $SUBDIR{"stride-mixed.dry-burial-10-7"} = "info";
    $SUBDIR{"stride-mixed.dry-burial-8-7"} = "info";
    $SUBDIR{"stride-mixed.dry-burial-6.5-7"} = "info";
    $SUBDIR{"stride-mixed.burial-12-7"} = "info";
    $SUBDIR{"stride-mixed.burial-10-7"} = "info";
    $SUBDIR{"stride-mixed.burial-9-7"} = "info";
    $SUBDIR{"stride-mixed.burial-8-7"} = "info";
    $SUBDIR{"stride-mixed.burial-6.5-7"} = "info";
    $SUBDIR{"stride-mixed.burial-6.5-9"} = "info";
    $SUBDIR{"stride-mixed.abs_sa8"} = "info";
    $SUBDIR{"stride-mixed.abs_sa7"} = "info";
    $SUBDIR{"stride-mixed.rel_sa2"} = "info";
    $SUBDIR{"stride-mixed.rel_sa3"} = "info";
    $SUBDIR{"stride-mixed.rel_sa7"} = "info";
    $SUBDIR{"stride-mixed.rel_sa10"} = "info";

    $SUBDIR{"stride-mixed.near-backbone-7"} = "info";
    $SUBDIR{"stride-mixed.near-backbone-11"} = "info";
    $SUBDIR{"stride-mixed.way-back-7"} = "info";
    $SUBDIR{"stride-mixed.way-back-11"} = "info";

#extracted properties files
   $SUBDIR{"dihedral.rdb"} = "info";
   $SUBDIR{"dssp.rdb"} = "info";
   $SUBDIR{"gen6.5.burial.rdb"} = "info";
   $SUBDIR{"gen8.burial.rdb"} = "info";
   $SUBDIR{"gen9.burial.rdb"} = "info";
   $SUBDIR{"gen10.burial.rdb"} = "info";
   $SUBDIR{"gen12.burial.rdb"} = "info";
   $SUBDIR{"wet6.5.burial.rdb"} = "info";
   $SUBDIR{"wet8.burial.rdb"} = "info";
   $SUBDIR{"wet10.burial.rdb"} = "info";
   $SUBDIR{"wet12.burial.rdb"} = "info";
   $SUBDIR{"dry6.5.burial.rdb"} = "info";
   $SUBDIR{"dry8.burial.rdb"} = "info";
   $SUBDIR{"dry10.burial.rdb"} = "info";
   $SUBDIR{"dry12.burial.rdb"} = "info";
   $SUBDIR{"CB6.5.burial.rdb"} = "info";
   $SUBDIR{"CB8.burial.rdb"} = "info";
   $SUBDIR{"CB10.burial.rdb"} = "info";
   $SUBDIR{"CB12.burial.rdb"} = "info";
   $SUBDIR{"CA6.5.burial.rdb"} = "info";
   $SUBDIR{"CA8.burial.rdb"} = "info";
   $SUBDIR{"CA10.burial.rdb"} = "info";
   $SUBDIR{"CA12.burial.rdb"} = "info";

#the master
   $SUBDIR{"pdb-rdb"} = "info";


    # rename the suffix for the following abbreviations
    
    
    $override_suffix{"guide"} = ".hssp.guide";
    $override_suffix{"pdb-rdb"} = "pdb-rdb";



# --------------------------------------------------
# Read ids from file into a hash table.
# Using a hash means that duplicate IDs are automatically merged.
sub ReadIDs($\%)
{
    my ($file, $href) = @_;
    open(FILE, "<$file")
    	|| fatal("Can't open file $file for reading IDs");
    
    while (<FILE>)
    {   next if /^\s*#/;	# # is comment
        next if /^\s*\/\//;	# // is also comment
	next if ( !(  	/^\s*(\S+)/ && ($id = $1)) );
	$$href{$id} = 1;
    }
    close FILE;
}

# Read ids from FileHandle into a hash table.
# Using a hash means that duplicate IDs are automatically merged.
sub ReadIDsFH(*\%)
{
    my ($fileref, $href) = @_;
    
    while (<$fileref>)
    {   next if /^\s*#/;	# # is comment
        next if /^\s*\/\//;	# // is also comment
	next if ( !(  	/^\s*(\S+)/ && ($id = $1)) );
	$$href{$id} = 1;
    }
}


# --------------------------------------------------
# Read ids from file into an array.
sub ReadIDsToArray($\@)
{
    my ($file, $aref) = @_;
    open(FILE, "<$file")
        || fatal("Can't open file $file for reading IDs");
    
    while (<FILE>)
    {   next if /^\s*#/;# # is comment
            next if /^\s*\/\//;# // is also comment
            next if ( !(  /^\s*(\S+)/ && ($id = $1)) );
        push @$aref, $id;
    }
    close FILE;
}


#print the keys of the hash (reference) to STDOUT
sub PrintIDsStdout(%) {
   use strict;
    my ($my_ids) = @_;
    my $key; 
   foreach $key (keys(%$my_ids)){
       print "$key\n";
    }
}


# return the subdirectory for a specified alignment
sub subdir_for_alignspec($)
{   my ($alignspec) = @_;
    if (! defined $SUBDIR{$alignspec})
    {
	print STDERR "ERROR: <alignment specifier> \"$alignspec\" unrecognized.\n";
	print STDERR "legal specifiers are:\n";
	foreach $x (sort keys(%SUBDIR))
	{    print STDERR "\t$x\n";
	}
	fatal("<alignment specifier> \"$alignspec\" unrecognized.\n");
    }
    return $SUBDIR{$alignspec};
}

    
# return the correct file suffix for the specified alignment spec
sub get_suffix($)
{   my ($spec) = @_;
    
    # append a ".a2m" unless a specific override has been defined,
    # or the alignment spec already contains a period.
    
    if ( defined($override_suffix{$spec}))
    {    return $override_suffix{$spec};
    }
    elsif ($spec =~ /[.]/)
    {    return $spec;
    }
    else
    {    return $spec . ".a2m";
    }

}


# Return the full path for the DIRECTORY containing all
# information about a specified id.
# UGLY--relies on $ChainIds::models97_subdir and $ChainIds::chain_subdir being set.
sub dir_path_for_id($)
{    my ($id) = @_;
     return "$models97/$models97_subdir/" . substr($id,0,2) 
		    . "/$id/$chain_subdir";
}


#version of above that has models97_subdir and chain_subdir passed
#in as arguments
sub pretty_dir_path_for_id($$$)
{    my ($id,$m97_subdir,$c_subdir) = @_;
     return "$models97/$m97_subdir/" . substr($id,0,2) 
		    . "/$id/$c_subdir";
}


# Construct the full filename for given id and alignment spec,
# searching array of possible subdirectories (usually just ("pdb") ).
# Return the full file name if file is found, otherwise return undef.
# Example: full_existing_filename("1kcqA", "stride.seq", "pdb")
#	returns /projects/compbio/experiments/models.97/pdb/1k/1kcqA/info/1kcqA.stride.seq

sub full_existing_filename($$@)
{   my ($id, $spec, @subdir_list) = @_;
    my $suffix = get_suffix($spec);
    my $chain_subdir = subdir_for_alignspec($spec);
    
    # We have to find the appropriate subdirectory:
    foreach my $subdir (@subdir_list)
    {   $full_subdir =  "$models97/$subdir/" . substr($id,0,2) 
		    . "/$id/$chain_subdir";
 	 $path_nogz = "$full_subdir/$id.$suffix";
	 
	 $path = $path_nogz . ".gz";
	 return $path if (-e $path);
	 $path = $path_nogz;
	 return $path if (-e $path);
     }
     
     return undef;

}

# Extends full_existing_filename to look for files in the
# /pcem/removed-templates directory
# Return the full file name if file is found, otherwise return undef.
sub full_existing_obsolete_filename($$@)
{   my ($id, $spec, @subdir_list) = @_;
    my $suffix = get_suffix($spec);
    my $chain_subdir = subdir_for_alignspec($spec);

    # We have to find the appropriate subdirectory:
    $full_subdir =  $models97 . "/removed-templates/" 
		    . "/$id/$chain_subdir";
    $path_nogz = "$full_subdir/$id.$suffix";
	 
    $path = $path_nogz . ".gz";
    return $path if (-e $path);
    $path = $path_nogz;
    return $path if (-e $path);
     
     
    return undef;

}


# calls pdb-get to get the full file name for a pdb chain
sub pdb_file($)
{   my ($id) = @_;
    my $file = `/projects/compbio/bin/pdb-get $id`;
    chomp $file;
    return $file
}


#
# fatal(msg)
#
# Generate a fatal error message.
#
sub fatal($) 
{   my($msg) = @_;
    die("*** Error: $PROGRAM_NAME: $msg\n");
    exit(1);
}

#
# run_prog(cmd, [errmsg])
#
# Run a command using system.  Exit if the command fails.  An optional
# error message for failure maybe specified or one will be generated.
#
sub run_prog($;$) 
{   my($cmd,$errmsg) = @_;
    if ($::verbose)
    {   print STDERR "\n\n@@@@ $cmd\n";
    }

    if (system($cmd) != 0) 
    {	if (!defined($errmsg)) 
        {   $errmsg = "command failed: $cmd";
        }
        fatal($errmsg);
    }
}

# packages must end with a true value---a restriction that seems
# to be poorly documented.
1;