small fixes/improvements

annot.nf

@@ -191,10 +191,12 @@ if (params.run_exonerate) {
           -join -type CDS -translate -retainids 1 > ref.pep
         """
     }
-    exn_prot_chunk = ref_pep.splitFasta( by: 20)
+    exn_prot_chunk = ref_pep.splitFasta( by: 100)
     exn_genome_chunk = pseudochr_seq_exonerate.splitFasta( by: 3)
     process run_exonerate {
         cache 'deep'
+        // this process can fail for rogue exonerate processes
+        errorStrategy 'ignore'
 
         input:
         set file('genome.fasta'), file('prot.fasta') from exn_genome_chunk.spread(exn_prot_chunk)

bin/iproscan_gff3_merge.lua

@@ -142,16 +142,25 @@ function annotate_vis:visit_feature(fn)
       for k,v in pairs(hits) do
         if not FILTERED_SOURCES[k] then
           for _,n in ipairs(v) do
-            rng = aminoloc_to_dnaloc(fn, n:get_range(), n:get_strand())
-            new_node = gt.feature_node_new(fn:get_seqid(), "protein_match",
-                                           rng[1], rng[2], fn:get_strand())
-            new_node:set_source(k)
-            for attr, attrv in n:attribute_pairs() do
-              if not FILTERED_ATTRIBS[attr] then
-                new_node:set_attribute(attr, string.gsub(attrv, "\"",""))
+            local rng = aminoloc_to_dnaloc(fn, n:get_range(), n:get_strand())
+            if fn:get_range():contains(gt.range_new(rng[1],rng[2])) then
+              local new_node = gt.feature_node_new(fn:get_seqid(),
+                                                   "protein_match",
+                                                   rng[1], rng[2],
+                                                   fn:get_strand())
+              new_node:set_source(k)
+              for attr, attrv in n:attribute_pairs() do
+                if not FILTERED_ATTRIBS[attr] then
+                  new_node:set_attribute(attr, string.gsub(attrv, "\"",""))
+                end
               end
+              fn:add_child(new_node)
+            else
+              io.stderr:write("coordinates for feature outside of parent: "
+                               .. tostring(fn:get_range()) .. " vs. "
+                               .. tostring(rng) .. " -- not attaching to "
+                               .. "polypeptide parent")
             end
-            fn:add_child(new_node)
           end
         end
       end

bin/sample_ref_genes.lua

@@ -20,7 +20,7 @@
 math.randomseed(os.time())
 
 function usage()
-  io.stderr:write("Randomly samples a number of protein coding gene CCs.\n")
+  io.stderr:write("Randomly samples a number of single transcript protein coding gene CCs.\n")
   io.stderr:write(string.format("Usage: %s <GFF with gene annotations> " ..
                                 "<number of genes to sample>\n" , arg[0]))
   os.exit(1)
@@ -36,17 +36,19 @@ cv.out = nil
 function cv:visit_feature(fn)
   local gene = false
   local mrna = false
+  local nof_transcripts = 0
   local cds = false
   for n in fn:get_children() do
     if n:get_type() == "gene" then
       gene = true
     elseif n:get_type() == "mRNA" then
       mrna = true
+      nof_transcripts = nof_transcripts + 1
     elseif n:get_type() == "CDS" then
       cds = true
     end
   end
-  if gene and mrna and cds then
+  if gene and mrna and nof_transcripts == 1 and cds then
     self.out = fn
   else
     self.out = nil

loc_sanger_farm.config

@@ -20,8 +20,12 @@ process.memory = "2 GB"
 process.clusterOptions = " -R 'select[mem>2048] rusage[mem=2048]' "
 process.$contiguate_pseudochromosomes.memory = "8 GB"
 process.$contiguate_pseudochromosomes.clusterOptions = " -R 'select[mem>8192] rusage[mem=8192]' "
+process.$blast_for_circos.memory = "8 GB"
+process.$blast_for_circos.clusterOptions = " -R 'select[mem>8192] rusage[mem=8192]' "
 process.$run_ratt.memory = "8 GB"
 process.$run_ratt.clusterOptions = " -R 'select[mem>8192] rusage[mem=8192]' "
+process.$run_exonerate.memory = "8 GB"
+process.$run_exonerate.clusterOptions = " -R 'select[mem>8192] rusage[mem=8192]' "
 process.$run_augustus_pseudo.memory = "2 GB"
 process.$run_augustus_pseudo.clusterOptions = " -R 'select[mem>2048] rusage[mem=2048]' "
 process.$predict_ncRNA.memory = "8 GB"