From 816d6b6dcc55515943046246e5f65382fd6fca0c Mon Sep 17 00:00:00 2001
From: Amit Lavon <amit.lavon@weizmann.ac.il>
Date: Sun, 8 Mar 2020 10:01:33 +0200
Subject: [PATCH] Speed up overlap lookup in CDS stage Also skip lookup if flag
 says to.

Conflicts:
	bin/prokka
---
 bin/prokka | 28 +++++++++++++++++++++-------
 1 file changed, 21 insertions(+), 7 deletions(-)

diff --git a/bin/prokka b/bin/prokka
index d92551e..381149a 100755
--- a/bin/prokka
+++ b/bin/prokka
@@ -709,6 +709,17 @@ if ($tools{'minced'}->{HAVE}) {
 # . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 
 # CDS
 
+# Create a hash of RNA lists, grouped by seqid to speed up lookups.
+my %rnabyseq;
+for my $rna (@allrna) {
+  my @a = ();
+  $rnabyseq{$rna->seq_id} = \@a;
+}
+for my $rna (@allrna) {
+  my $a = $rnabyseq{$rna->seq_id};
+  push @$a, $rna;
+}
+
 msg("Predicting coding sequences");
 my $totalbp = sum( map { $seq{$_}{DNA}->length } @seq);
 my $prodigal_mode = ($totalbp >= 100000 && !$metagenome) ? 'single' : 'meta';
@@ -744,15 +755,18 @@ while (<$PRODIGAL>) {
       }
     );
     my $overlap;
-    for my $rna (@allrna) {
-      # same contig, overlapping (could check same strand too? not sure)
-      if ($rna->seq_id eq $sid and $cds->overlaps($rna)) { 
-        $overlap = $rna;
-	last;
-      }	
+    if (!$cds_rna_olap and exists($rnabyseq{$sid})) {
+      my $seqrna = $rnabyseq{$sid};
+      for my $rna (@$seqrna) {
+        # same contig, overlapping (could check same strand too? not sure)
+        if ($rna->seq_id eq $sid and $cds->overlaps($rna)) {
+          $overlap = $rna;
+	        last;
+        }	
+      }
     }
     # mitochondria are highly packed, so don't exclude as CDS/tRNA often overlap.
-    if ($overlap and ! $cds_rna_olap) {
+    if ($overlap) {
       my $type = $overlap->primary_tag;
       msg("Excluding CDS which overlaps existing RNA ($type) at $sid:$1..$2 on $3 strand");
     }