biotree: add --boot-all; bioaln: add orf-slice

Author: wqiu

bin/bioaln

@@ -416,17 +416,18 @@ Generate an alignment of every-third (mostly synonymous) bases (assuming a CDS a
 
 Make a shuffled (not bootstrapped, which is sampling I<with> replacement) alignment. This operation I<permutes> alignment columns. It is used for testing the significance of long-runs of conserved sites in an alignment (e.g., conserved intergenic spacer sequences).
 
-=item --slice, -s 'start|-,end|-'
+=item --slice, -s 'start|-,end|-'; or -s"file:orfs.tsv"
 
 Get a slice of the alignment.
 
 Using a '-' character in the first or second position defaults to the beginning or end, respectively. Therefore specifying -s'-,-' is the same as grabbing the whole alignment.
 
- --slice'20,80' or --slice '20,80' or -s='20,80' or --slice='20,80': Slice from position 20 to 80, inclusive.
- --slice'-,80':  Slice from beginning up to, and including position 80
- --slice'20,-':  Slice from position 20 up to, and including, the end of the alignment
+ --slice '20,80' or --slice '20,80' or -s='20,80' or --slice='20,80': Slice from position 20 to 80, inclusive.
+ --slice '-,80':  Slice from beginning up to, and including position 80
+ --slice '20,-':  Slice from position 20 up to, and including, the end of the alignment
+ --slice 'file:orfs.tsv': slice using coordinates from a file. The file should contain 4 tab/space-delimited columns: locus_tag (no space), start (numerical), end (numerical), strand(0/1)
 
-NOTE: --slice'-,x' (where x is '-' or a position) does NOT work. Use --slice='-,x' (or a space in place of =) instead.
+NOTE: --slice'-,x' (where x is '-' or a position) without a space does NOT work. Use --slice='-,x' (or a space in place of =) instead.
 
 =item --split-cdhit <cdhit clrs file>
 

bin/biotree

@@ -21,6 +21,7 @@ GetOptions(\%opts,
 	   "man",
 	   "version|V",
 	   "as-text|t", # for preview text tree
+	   "boot-all",  # show support values
 	   "ci|c=s", # attach binary trait table
 	   "clean-br|b",
 	   "clean-boot|B",
@@ -123,6 +124,10 @@ B<biotree> supports reading from STDIN, so that multiple tree manipulations coul
 
 Draw an ASCII tree for quick preview (needs refinement). Default max screen width 100 characters.
 
+=item --boot-all
+
+Show support values for all nodes
+
 =item --ci, -c 'binary-trait-file'
 
 Attach a file containing binary trait values and prints consistency index for informative sites (not verified)

lib/Bio/BPWrapper/AlnManipulations.pm

@@ -809,13 +809,37 @@ with improvements.
 
 
 sub aln_slice {    # get alignment slice
-    my ($begin, $end) = split(/\s*,\s*/, $opts{"slice"});
-
+    my $opt_str = $opts{"slice"};
+    my $id;
+    my $begin;
+    my $end;
+    my $strand;
+    if ($opt_str =~ /^file:(\S+)$/) {
+	my $fname = $1;
+	open COORD, "<", $fname;
+	while(<COORD>) {
+	    chomp;
+	    ($id, $begin, $end, $strand) = split;
+	}
+    } else {
+	($begin, $end) = split(/\s*,\s*/, $opt_str);
+    }
+    
     # Allow for one parameter to be omitted. Default $begin to the
     # beginning of the alignment, and $end to the end.
     $begin = 1            if $begin eq "-";
     $end   = $aln->length if $end   eq "-";
-    $aln = $aln->slice($begin, $end)
+    $aln = $aln->slice($begin, $end);
+    if ($strand && $strand == 0) {
+	my $new_aln = Bio::SimpleAlign -> new();
+	foreach ($aln->each_seq) {
+	    my $revcom = $_ -> revcom();
+	    my $end = $_ -> end;
+	    $revcom->end($end);
+	    $new_aln->add_seq($revcom);
+	}
+	$aln = $new_aln;
+    }
 }
 
 =head2 get_unique()

lib/Bio/BPWrapper/TreeManipulations.pm

@@ -1188,6 +1188,7 @@ Call this after calling C<#initialize(\%opts)>.
 
 sub write_out {
     my $opts = shift;
+    print_all_boots() if $opts->{'boot-all'};
     print_all_node_ids() if $opts->{'ids-all'};
     rename_tips() if $opts->{'rename-tips'};
     write_tab_tree() if $opts->{'as-text'};
@@ -1337,6 +1338,17 @@ sub print_all_node_ids {
     }
 }
 
+
+sub print_all_boots {
+    my %visited;
+    my @node_list = ($rootnode);
+    &_wu($rootnode, \%visited, \@node_list);
+    foreach (@node_list) {
+	next if $_->is_Leaf || $_ eq $rootnode;
+	print $_->internal_id(), "\t", $_->id() || "NA", "\n";
+    }
+}
+
 sub _treeheight {
 	my $height = 0;
 	my $tree = $_[0];