#!/usr/bin/perl -w

#
# Chris Dragon, 9-9-03: Added comments.  I'm not an expert on these systems,
# so my comments are only my best guess as to what various things are doing.
# If you see a comment that makes no sense to you, it could very well be wrong.
# I'll prefix my comments with CD:.
#
# CD: Generates an HTML summary report containing the best alignments for a
# sequence.
#
# Usage: 
#	casp_summary_report_html \
#		--align target.top_reported_alignments.rdb \
#		--target target \
#		[--make_al] \
#		> casp_summary_report.html
#
# CD:
# --align specifies the RDB file containing the best alignments chosen for
#   display.
# --target contains the sequence ID (ie. 1edn) of the target sequence.
# --make_al If true, we'll generate .al files for each .a2m file in the RDB
#   file specified by --align.
#
# Update history:
#   Chris Dragon, 9-10-03: Copies .al files to MODEL[model num].al files.

sub print_data_row($$$$$$$);
sub find_al_file($$);
sub make_al_file($$);
sub read_rdb_header(*\@\%);

{
    use FileHandle;
    STDERR->autoflush(1);
    use English;
    use File::Basename;
    use File::stat;
    use Getopt::Long;

    $modnum=0;		# model number for .al format files
    $conversion_prog = dirname($PROGRAM_NAME)."/a2m-to-caspal.perl";
    
    $make_al_alignments = 0;

    # Get command line options using the Getopt::Long standard perl module.
    GetOptions("align=s" => \$predict_aligns_rdb,
	"target=s" => \$TARGET,
	"make_al" => \$make_al_alignments);
	
    if ($make_al_alignments)
    {	print STDERR "# Will make .al format for alignments\n";
    }
    
    if (!defined($predict_aligns_rdb)
    	|| !defined($TARGET))
    {	&print_usage_exit; 
    }


    # foo-predicted_alignments.rdb contains 5 fields 
    #Alignment	Evalue	Align_Evalue	FSSP	SCOP
    #50S	10N	10N	5S	50S

    #We want an html page with a table which tells us for each top alignment:
    # - link to a2m formatted alignment
    # - alignment E-value
    # fssp rep of template (plus link)
    # all SCOP domains found in the template (plus link)



    open(PREDICT_ALIGNS, "$predict_aligns_rdb") || die "Can't open $predict_aligns_rdb for reading\n";
    my @colname;	# names of columns in rdb file
    my %col_num;		# mapping name->number
    read_rdb_header(*PREDICT_ALIGNS{IO},@colname, %col_num);
	$aligncolnum = $col_num{"Alignment"};
	$Ecolnum = $col_num{"Evalue"};
	$AEcolnum = $col_num{"Align_Evalue"};
    	$FSSPcolnum = $col_num{"FSSP"};
    	$SCOPcolnum = $col_num{"SCOP"};
    	$SUIDcolnum = $col_num{"SUID"};

    &begin_html_table;

    while ($line = <PREDICT_ALIGNS>)
    {   chomp $line;
	@cols = split("\t", $line);
	#get template name from alignment name
	
        $a2m_name=$cols[$aligncolnum];
	$a2m_name =~ /^([^\/]+)\//;
        $templateid = $1;

	if ($make_al_alignments)
	{   $alfilename = $a2m_name;
	    $alfilename =~ s/.a2m/.al/;
	    find_al_file($alfilename,$a2m_name);
	}
	else 
	{   $alfilename = "";
	}

	&print_data_row($cols[$aligncolnum], $alfilename, $cols[$Ecolnum],
			$cols[$AEcolnum], $cols[$FSSPcolnum],
			$cols[$SCOPcolnum], $cols[$SUIDcolnum]);
    }
    close(PREDICT_ALIGNS);
    &end_html_table;
}

sub print_data_row($$$$$$$)
{   ($a2mfile, $alfile, $best_evalue, $align_evalue, $fssprep, $scop, $suid ) = @_;
    print "<tr>\n";
    print "<td><a href=\"$a2mfile\">$a2mfile</a></td>\n";
    print "<td><a href=\"$alfile\">$alfile</a></TD>\n" if ($make_al_alignments);
    print "<td>$best_evalue</td><td>$align_evalue</td>\n";
    print "<td>";
    
    if(defined $fssprep)
    {   # KNOWN BUG: FSSP NO LONGER EXISTS---need to update to newer scheme
    	# but I was unable to download the Dali Domain Dictionary on 9 May 2005  
    	print "<a href=\"http://jura.ebi.ac.uk:8765/holm/qz?$fssprep\">$fssprep</a>";
    }
    print "</td>\n";
    print "<td>";
    if (defined($scop))
    {	my @scopdoms = split(',', $scop);
	my @scopsuids = split(',', $suid);
	for (my $i =0; $i < scalar(@scopdoms); $i++)
	{   print "<BR>\n" if ($i>0);
	    print "<a href=\"http://scop.mrc-lmb.cam.ac.uk/scop/search.cgi?sunid=$scopsuids[$i]\">$scopdoms[$i]</a>";
	} 
    }
    print "</td>\n";
    print "</tr>\n";
}

sub begin_html_table
{
    $title = "Best Alignments for $TARGET";

    print  "<HTML>\n<HEAD>\n";
    print  "<!-- this file was generated by the casp_summary_report_html script -->";
    print  "<TITLE>$title</TITLE></HEAD>\n";
    print "\n<BODY \n";
    print "background=\"http://www.soe.ucsc.edu/projects/compbio/.html/backgrounds/1o6d-pale.gif\" ";
    print " >\n";
    print  "<CENTER><H1>$title</H1>\n";
    print "</CENTER>\n";

    print "To select the top models, SAM generates pairwise\n";
    print "alignments of the target sequence and the best-scoring\n";
    print "templates.  The alignments with the best E-values are\n";
    print "presented here.\n";
    print "<P>\n";
    print "Warning: E-values smaller than about 0.001 may be overly optimistic.\n";
    print "<P>\n";

    $align_width = ($make_al_alignments? 30: 60);
    
    print  "<CENTER>\n";
    #start the table and print header row
    print "<table width=\"100%\" border cellspacing=\"0\" cellpadding=\"0\">\n";
    print "<tr>\n";
    print "<td width=\"$align_width%\">A2M alignment</TD>\n";
    if ($make_al_alignments)
    {   print "<td width=\"$align_width%\">AL alignment</TD>\n";
    }
    print "<td width=\"10%\">Template E-value</TD>\n";
    print "<td width=\"10%\">Alignment cost</TD>\n";
    print "<td width=\"10%\">FSSP rep</TD>\n";
    print "<td width=\"10%\">SCOP Domain(s)</TD>\n";
    print "</tr>\n";
}

sub end_html_table()
{
    #finish the table
    print "</table>\n";

    #finish off the page with a citation
    print "</CENTER>\n";
    print "<P><P>\n";
    
    print "Please cite: Karplus, K. and Karchin, R. and Barrett, C. and Tu, S. and Cline, M. and Diekhans, M. and Grate, L. and Casper, J. and Hughey, R. ``What is the value added by human intervention in protein structure prediction?''\n"; 
    print "<I>Proteins: Structure Function and Genetics </I>45(S5):86-91,2001\n";
    print "</BODY></HTML>\n\n";
}

sub print_usage_exit()
{    print "Usage:	casp_summary_report_html \
		--align target.predicted_alignments.rdb \
		--best target.best_scores.rdb \
		--target target \
		[--make_al] \
		> casp_summary_report.html\n";
    exit(-1);
}

sub find_al_file($$)
{   my ($alfilename,$a2m_name) = @_;
    print STDERR "# looking for $alfilename\n";
    if ( ! -e "$alfilename")
    {	# the .al file doesn't exist, so make it
    	make_al_file($alfilename,$a2m_name);
    }
    else 
    {    # the file exists, but is it up to date?
	# Need to do stat($alfilename) and stat($a2m_name) and compare
	#	mtime s
	my $alstat = stat($alfilename);
	my $a2mstat= stat($a2m_name);
	if ($alstat->mtime < $a2mstat->mtime)
	{   # oops, .al file is older, remake it
	    make_al_file($alfilename,$a2m_name);
	}    
    }
}

sub make_al_file($$)
{   my ($alfilename,$a2m_name) = @_;
    my ($SAM_YEAR) = '05';
    
    print STDERR "Will make $alfilename\n";

    #a2m-to-caspal.perl takes an argument file that we must produce here

    #-infasta in1.a2m
    #-outcaspal out1.casp.al
    #-target 1EKT:B TRANSCRIPTION STATE REGULATORY PROTEIN ABRB
    #-template 1ev13 Chain 3, Echovirus 1
    #-author UCSC SAM-T${SAM_YEAR} server
    #-method method.txt

    (defined $author) or $author = "UCSC SAM-T${SAM_YEAR} server";

    $modnum++;
    $alignEvalue = $cols[$Ecolnum];
    $argfile = $alfilename;
    $argfile =~ s/[.]al$/.args/;
    open(ARGFILE, ">$argfile") || die "Can't open $argfile for writing.\n";
    print ARGFILE "-infasta $a2m_name\n";
    print ARGFILE "-outcaspal $alfilename\n";
    print ARGFILE "-target $TARGET\n";
    print ARGFILE "-template $templateid\n";
    print ARGFILE "-author $author\n";
    if (defined($method))
    {  print ARGFILE "-method $method\n";
    }
    print ARGFILE "-modelnum $modnum\n";
    print ARGFILE "-score $alignEvalue\n";
    close(ARGFILE);

    system("$conversion_prog $argfile");
    unlink $argfile;
}

# read in the header of an RDB file and store the column names in @$col_names_ref
# with a reversed index in %$col_num_ref.
sub read_rdb_header(*\@\%)
{   my ($FILE,$col_names_ref,$col_num_ref) = @_;
    # skip the comments at the beginning
    while (<$FILE> )
    {	last if /.+/ && !(/^\s*#/);
    }
    
    # read the column names
    chomp;
    @$col_names_ref = split(/\t/);

    # index the column names
    for(my $i = 0; $i<scalar(@$col_names_ref); $i++)
    {   $$col_num_ref{$$col_names_ref[$i]} = $i;
        # print STDERR "column $i is $$col_names_ref[$i]\n";
    }
    
    # skip the field-length definitions
    <$FILE>;
}


# CHANGE LOG: 
#  27 Sept 2003 Kevin Karplus
#	Modified find_al_file and moved code to make_al_file
#	so that al files that were older than a2m files would be remade.
#
#   Sun May  8 10:07:24 PDT 2005 Kevin Karplus
#	Eliminated reading best-scores, since all info now in the alignments rdb file