Parent Directory
| 
Revision Log
Revision 1.23 - (view) (download) (as text)
| 1 : | efrank | 1.1 | # -*- perl -*- |
| 2 : | olson | 1.20 | # |
| 3 : | # Copyright (c) 2003-2006 University of Chicago and Fellowship | ||
| 4 : | # for Interpretations of Genomes. All Rights Reserved. | ||
| 5 : | # | ||
| 6 : | # This file is part of the SEED Toolkit. | ||
| 7 : | # | ||
| 8 : | # The SEED Toolkit is free software. You can redistribute | ||
| 9 : | # it and/or modify it under the terms of the SEED Toolkit | ||
| 10 : | # Public License. | ||
| 11 : | # | ||
| 12 : | # You should have received a copy of the SEED Toolkit Public License | ||
| 13 : | # along with this program; if not write to the University of Chicago | ||
| 14 : | # at info@ci.uchicago.edu or the Fellowship for Interpretation of | ||
| 15 : | # Genomes at veronika@thefig.info or download a copy from | ||
| 16 : | # http://www.theseed.org/LICENSE.TXT. | ||
| 17 : | # | ||
| 18 : | |||
| 19 : | efrank | 1.1 | |
| 20 : | use strict; | ||
| 21 : | use FIG; | ||
| 22 : | my $fig = new FIG; | ||
| 23 : | |||
| 24 : | parrello | 1.18 | use Tracer; |
| 25 : | overbeek | 1.21 | if ($ENV{'VERBOSE'}) { TSetup("3 FIG DBKernel","TEXT") } |
| 26 : | efrank | 1.1 | |
| 27 : | parrello | 1.18 | Trace("Preparing to load features.") if T(2); |
| 28 : | my ($mode, @genomes) = FIG::parse_genome_args(@ARGV); | ||
| 29 : | efrank | 1.1 | my $temp_dir = "$FIG_Config::temp"; |
| 30 : | parrello | 1.18 | my $organisms_dir = "$FIG_Config::organisms"; |
| 31 : | efrank | 1.1 | |
| 32 : | my($genome,@types,$type,$id,$loc,@aliases,$aliases,$contig); | ||
| 33 : | |||
| 34 : | # usage: load_features [G1 G2 G3 ... ] | ||
| 35 : | |||
| 36 : | parrello | 1.18 | Open(\*REL, ">$temp_dir/tmpfeat$$"); |
| 37 : | Open(\*ALIAS, "| sort -T $temp_dir -u > $temp_dir/tmpalias$$"); | ||
| 38 : | overbeek | 1.21 | Open(\*DELFIDS,"| sort -u > $temp_dir/tmpdel$$"); |
| 39 : | Open(\*REPFIDS,"| sort -u > $temp_dir/tmprel$$"); | ||
| 40 : | |||
| 41 : | if ($mode eq 'all') { | ||
| 42 : | |||
| 43 : | overbeek | 1.22 | # Process any remaining deleted.features or replaced.features in Global |
| 44 : | if (open(GLOBDEL,"<$FIG_Config::global/deleted.features")) | ||
| 45 : | overbeek | 1.21 | { |
| 46 : | while (defined($_ = <GLOBDEL>) && ($_ =~ /^fig\|(\d+\.\d+)/)) | ||
| 47 : | { | ||
| 48 : | print DELFIDS "$1\t$_"; | ||
| 49 : | } | ||
| 50 : | } | ||
| 51 : | close(GLOBDEL); | ||
| 52 : | |||
| 53 : | overbeek | 1.22 | if (open(GLOBREP,"<$FIG_Config::global/replaced.features")) |
| 54 : | overbeek | 1.21 | { |
| 55 : | while (defined($_ = <GLOBREP>) && ($_ =~ /^fig\|(\d+\.\d+)/)) | ||
| 56 : | { | ||
| 57 : | print REPFIDS "$1\t$_"; | ||
| 58 : | } | ||
| 59 : | } | ||
| 60 : | close(GLOBREP); | ||
| 61 : | efrank | 1.1 | |
| 62 : | |||
| 63 : | overbeek | 1.10 | |
| 64 : | # Here we extract external aliases from the peg.synonyms table, when they can be inferred | ||
| 65 : | # accurately. | ||
| 66 : | parrello | 1.18 | Trace("Extracting external aliases from the peg.synonyms table.") if T(2); |
| 67 : | overbeek | 1.21 | Open(\*SYN, "<$FIG_Config::global/peg.synonyms"); |
| 68 : | overbeek | 1.10 | while (defined($_ = <SYN>)) |
| 69 : | { | ||
| 70 : | parrello | 1.18 | chop; |
| 71 : | my($x,$y) = split(/\t/,$_); | ||
| 72 : | my @ids = map { $_ =~ /^([^,]+),(\d+)/; [$1,$2] } ($x,split(/;/,$y)); | ||
| 73 : | my @fig = (); | ||
| 74 : | my(@nonfig) = (); | ||
| 75 : | foreach $_ (@ids) | ||
| 76 : | { | ||
| 77 : | if ($_->[0] =~ /^fig\|/) | ||
| 78 : | { | ||
| 79 : | push(@fig,$_); | ||
| 80 : | } | ||
| 81 : | else | ||
| 82 : | { | ||
| 83 : | push(@nonfig,$_); | ||
| 84 : | } | ||
| 85 : | } | ||
| 86 : | |||
| 87 : | foreach $x (@fig) | ||
| 88 : | overbeek | 1.13 | { |
| 89 : | parrello | 1.18 | my($peg,$peg_ln) = @$x; |
| 90 : | my $genome = &FIG::genome_of($peg); | ||
| 91 : | foreach $_ (@nonfig) | ||
| 92 : | { | ||
| 93 : | if ((@fig == 1) || ($peg_ln == $_->[1])) | ||
| 94 : | { | ||
| 95 : | print ALIAS "$peg\t$_->[0]\t$genome\n"; | ||
| 96 : | Trace("Alias record $peg, $_->[0] for $genome.") if T(4); | ||
| 97 : | } | ||
| 98 : | } | ||
| 99 : | overbeek | 1.13 | } |
| 100 : | overbeek | 1.10 | } |
| 101 : | close(SYN); | ||
| 102 : | efrank | 1.1 | } |
| 103 : | |||
| 104 : | foreach $genome (@genomes) | ||
| 105 : | { | ||
| 106 : | parrello | 1.18 | Trace("Processing $genome.") if T(3); |
| 107 : | efrank | 1.1 | opendir(FEAT,"$organisms_dir/$genome/Features") |
| 108 : | parrello | 1.18 | || die "could not open $genome/Features"; |
| 109 : | efrank | 1.1 | @types = grep { $_ =~ /^[a-zA-Z]+$/ } readdir(FEAT); |
| 110 : | closedir(FEAT); | ||
| 111 : | |||
| 112 : | foreach $type (@types) | ||
| 113 : | { | ||
| 114 : | overbeek | 1.22 | if ((-s "$organisms_dir/$genome/Features/$type/deleted.features") && |
| 115 : | open(TMP,"<$organisms_dir/$genome/Features/$type/deleted.features")) | ||
| 116 : | overbeek | 1.21 | { |
| 117 : | while (defined($_ = <TMP>) && ($_ =~ /^fig\|(\d+\.\d+)/)) | ||
| 118 : | { | ||
| 119 : | print DELFIDS "$1\t$_"; | ||
| 120 : | } | ||
| 121 : | close(TMP); | ||
| 122 : | } | ||
| 123 : | |||
| 124 : | overbeek | 1.22 | if ((-s "$organisms_dir/$genome/Features/$type/replaced.features") && |
| 125 : | open(TMP,"<$organisms_dir/$genome/Features/$type/replaced.features")) | ||
| 126 : | overbeek | 1.21 | { |
| 127 : | while (defined($_ = <TMP>) && ($_ =~ /^fig\|(\d+\.\d+)/)) | ||
| 128 : | { | ||
| 129 : | print REPFIDS "$1\t$_"; | ||
| 130 : | } | ||
| 131 : | close(TMP); | ||
| 132 : | } | ||
| 133 : | |||
| 134 : | overbeek | 1.19 | if ((-s "$organisms_dir/$genome/Features/$type/tbl") && |
| 135 : | open(TBL,"<$organisms_dir/$genome/Features/$type/tbl")) | ||
| 136 : | { | ||
| 137 : | Trace("Loading $genome/Features/$type/tbl") if T(4); | ||
| 138 : | my @tbl = <TBL>; | ||
| 139 : | close(TBL); | ||
| 140 : | my %seen; | ||
| 141 : | |||
| 142 : | while ($_ = pop @tbl) | ||
| 143 : | { | ||
| 144 : | chop; | ||
| 145 : | ($id,$loc,@aliases) = split(/\t/,$_); | ||
| 146 : | |||
| 147 : | if ($id && (! $seen{$id})) | ||
| 148 : | overbeek | 1.12 | { |
| 149 : | overbeek | 1.19 | $seen{$id} = 1; |
| 150 : | my($minloc,$maxloc); | ||
| 151 : | if ($loc) | ||
| 152 : | { | ||
| 153 : | $loc =~ s/\s+$//; | ||
| 154 : | parrello | 1.23 | ($contig,$minloc,$maxloc) = $fig->boundaries_of($loc); |
| 155 : | overbeek | 1.19 | if ($minloc && $maxloc) |
| 156 : | overbeek | 1.8 | { |
| 157 : | overbeek | 1.19 | ($minloc < $maxloc) || (($minloc,$maxloc) = ($maxloc,$minloc)); |
| 158 : | } | ||
| 159 : | } | ||
| 160 : | parrello | 1.18 | |
| 161 : | overbeek | 1.19 | if (! $contig) |
| 162 : | { | ||
| 163 : | $loc = $contig = $minloc = $maxloc = ""; | ||
| 164 : | } | ||
| 165 : | parrello | 1.18 | |
| 166 : | overbeek | 1.19 | if (@aliases > 0) |
| 167 : | { | ||
| 168 : | $aliases = join(",",grep(/\S/,@aliases)); | ||
| 169 : | my $alias; | ||
| 170 : | foreach $alias (@aliases) | ||
| 171 : | { | ||
| 172 : | if ($alias =~ /^([NXYZA]P_|gi\||sp\|\tr\||kegg\||uni\|)/) | ||
| 173 : | { | ||
| 174 : | |||
| 175 : | print ALIAS "$id\t$alias\t$genome\tOVERRIDE\n"; | ||
| 176 : | Trace("$id override alias $alias for $genome") if T(4); | ||
| 177 : | } | ||
| 178 : | overbeek | 1.8 | } |
| 179 : | overbeek | 1.19 | } |
| 180 : | else | ||
| 181 : | { | ||
| 182 : | $aliases = ""; | ||
| 183 : | } | ||
| 184 : | $minloc = (! $minloc) ? 0 : $minloc; | ||
| 185 : | $maxloc = (! $maxloc) ? 0 : $maxloc; | ||
| 186 : | if ((length($loc) < 5000) && (length($contig) < 96) && (length($id) < 32) && ($id =~ /(\d+)$/)) | ||
| 187 : | { | ||
| 188 : | print REL "$id\t$1\t$type\t$genome\t$loc\t$contig\t$minloc\t$maxloc\t$aliases\n"; | ||
| 189 : | } | ||
| 190 : | efrank | 1.1 | } |
| 191 : | overbeek | 1.19 | } |
| 192 : | } | ||
| 193 : | efrank | 1.1 | } |
| 194 : | } | ||
| 195 : | overbeek | 1.21 | close(REPFIDS); |
| 196 : | close(DELFIDS); | ||
| 197 : | efrank | 1.1 | close(REL); |
| 198 : | overbeek | 1.8 | close(ALIAS); |
| 199 : | parrello | 1.18 | Open(\*ALIASIN, "<$temp_dir/tmpalias$$"); |
| 200 : | Open(\*ALIASOUT, ">$temp_dir/tmpalias$$.1"); | ||
| 201 : | Trace("Parsing alias file.") if T(2); | ||
| 202 : | overbeek | 1.14 | $_ = <ALIASIN>; |
| 203 : | while ($_ && ($_ =~ /^(\S+)/)) | ||
| 204 : | { | ||
| 205 : | my @aliases = (); | ||
| 206 : | my $curr = $1; | ||
| 207 : | while ($_ && ($_ =~ /^(\S+)\t(\S+)(\t(\S+))?/) && ($1 eq $curr)) | ||
| 208 : | { | ||
| 209 : | parrello | 1.18 | push(@aliases,[$2,$3 ? 1 : 0]); |
| 210 : | $_ = <ALIASIN>; | ||
| 211 : | overbeek | 1.14 | } |
| 212 : | my $x; | ||
| 213 : | my $genome = &FIG::genome_of($curr); | ||
| 214 : | foreach $x (@aliases) | ||
| 215 : | { | ||
| 216 : | overbeek | 1.19 | if ($x->[1]) |
| 217 : | { | ||
| 218 : | print ALIASOUT "$curr\t$x->[0]\t$genome\n"; | ||
| 219 : | } | ||
| 220 : | else | ||
| 221 : | { | ||
| 222 : | my $i; | ||
| 223 : | for ($i=0; ($i < @aliases) && ((! $aliases[$i]->[1]) || (! &same_class($x->[0],$aliases[$i]->[0]))); $i++) {} | ||
| 224 : | if ($i == @aliases) | ||
| 225 : | { | ||
| 226 : | print ALIASOUT "$curr\t$x->[0]\t$genome\n"; | ||
| 227 : | } | ||
| 228 : | } | ||
| 229 : | overbeek | 1.14 | } |
| 230 : | } | ||
| 231 : | close(ALIASIN); | ||
| 232 : | close(ALIASOUT); | ||
| 233 : | unlink("$temp_dir/tmpalias$$"); | ||
| 234 : | efrank | 1.1 | |
| 235 : | overbeek | 1.21 | $fig->reload_table($mode, 'deleted_fids',"genome varchar(16), fid varchar(32)", |
| 236 : | { deleted_fids_fid_ix => 'fid', deleted_fids_genome_ix => 'genome' }, | ||
| 237 : | "$temp_dir/tmpdel$$",\@genomes); | ||
| 238 : | |||
| 239 : | unlink("$temp_dir/tmpdel$$"); | ||
| 240 : | |||
| 241 : | $fig->reload_table($mode, 'replaced_fids',"genome varchar(16), from_fid varchar(32), to_fid varchar(32)", | ||
| 242 : | { replaced_fids_from_ix => 'from_fid', | ||
| 243 : | replaced_fids_to_ix => 'to_fid', | ||
| 244 : | replaced_fids_genome_ix => 'genome' | ||
| 245 : | }, | ||
| 246 : | "$temp_dir/tmprep$$",\@genomes); | ||
| 247 : | |||
| 248 : | unlink("$temp_dir/tmprep$$"); | ||
| 249 : | |||
| 250 : | parrello | 1.18 | $fig->reload_table($mode, 'features', |
| 251 : | "id varchar(32), idN INTEGER, type varchar(16),genome varchar(16)," . | ||
| 252 : | "location TEXT," . | ||
| 253 : | "contig varchar(96), minloc INTEGER, maxloc INTEGER," . | ||
| 254 : | "aliases TEXT", | ||
| 255 : | { features_id_ix => "id", features_org_ix => "genome", | ||
| 256 : | features_type_ix => "type", features_beg_ix => "genome, contig, minloc" }, | ||
| 257 : | "$temp_dir/tmpfeat$$", \@genomes); | ||
| 258 : | unlink("$temp_dir/tmpfeat$$"); | ||
| 259 : | efrank | 1.1 | |
| 260 : | parrello | 1.18 | $fig->reload_table($mode, 'ext_alias', |
| 261 : | "id varchar(32), alias varchar(32), genome varchar(16)", | ||
| 262 : | { ext_alias_alias_ix => "alias", ext_alias_genome_ix => "genome", | ||
| 263 : | ext_alias_ix_id => "id" }, | ||
| 264 : | "$temp_dir/tmpalias$$.1", \@genomes ); | ||
| 265 : | overbeek | 1.8 | |
| 266 : | overbeek | 1.14 | unlink("$temp_dir/tmpalias$$.1"); |
| 267 : | parrello | 1.18 | Trace("Features loaded.") if T(2); |
| 268 : | overbeek | 1.14 | |
| 269 : | sub same_class { | ||
| 270 : | my($x,$y) = @_; | ||
| 271 : | |||
| 272 : | my $class1 = &classA($x); | ||
| 273 : | my $class2 = &classA($y); | ||
| 274 : | return ($class1 && ($class1 eq $class2)); | ||
| 275 : | } | ||
| 276 : | |||
| 277 : | sub classA { | ||
| 278 : | my($alias) = @_; | ||
| 279 : | |||
| 280 : | if ($alias =~ /^([^\|]+)\|/) | ||
| 281 : | { | ||
| 282 : | parrello | 1.18 | return $1; |
| 283 : | overbeek | 1.14 | } |
| 284 : | elsif ($alias =~ /^[NXYZA]P_[0-9\.]+$/) | ||
| 285 : | { | ||
| 286 : | parrello | 1.18 | return "refseq"; |
| 287 : | overbeek | 1.14 | } |
| 288 : | else | ||
| 289 : | { | ||
| 290 : | parrello | 1.18 | return ""; |
| 291 : | overbeek | 1.14 | } |
| 292 : | } |
| MCS Webmaster | ViewVC Help |
| Powered by ViewVC 1.0.3 |