Merge pull request #34 from satta/work

Latest work

.gitattributes

@@ -1,3 +1,5 @@
 data/** linguist-vendored
+RATT/** linguist-vendored
+ABACAS2/** linguist-vendored
 example-data/** linguist-vendored
 *.nf linguist-language=Groovy

annot.nf

@@ -1356,16 +1356,17 @@ process make_report {
     input:
     set file('pseudo.fasta.gz'), file('scaf.fasta.gz') from report_inseq
     set file('pseudo.gff3'), file('scaf.gff3') from report_gff3
+    val params.SPECK_TEMPLATE
     val specfile
 
     output:
     set file('pseudo.report.html'), file('scaf.report.html') into report_output
 
     """
     gt speck -specfile ${specfile} -matchdescstart -seqfile pseudo.fasta.gz \
-      -provideindex -typecheck so -output html pseudo.gff3 > pseudo.report.html
+      -provideindex -typecheck so -output "${params.SPECK_TEMPLATE}" pseudo.gff3 > pseudo.report.html
     gt speck -specfile ${specfile} -matchdescstart -seqfile scaf.fasta.gz \
-      -provideindex -typecheck so -output html scaf.gff3 > scaf.report.html
+      -provideindex -typecheck so -output "${params.SPECK_TEMPLATE}" scaf.gff3 > scaf.report.html
     """
 }
 

bin/infernal_to_gff3.lua

@@ -20,7 +20,7 @@
 -- TODO: make this dynamic
 models = {}
 models.rRNA = {"SSU_rRNA_eukarya","5S_rRNA", "5_8S_rRNA"}
-models.snRNA = {"U2", "U6"}
+models.snRNA = {"U1", "U2", "U4", "U5", "U6"}
 models.snoRNA = {"snoTBR5", "snoTBR17", "snoTBR7"}
 
 package.path = gt.script_dir .. "/?.lua;" .. package.path

bin/pseudo_merge_with_genes.lua

@@ -244,7 +244,7 @@ function stream:process_current_cluster()
             io.stderr:write("partial alignment ".. tostring(best) ..
                             " ignored as it is not degenerated\n")
           end
-          table.insert(self.outqueue, deep_copy(best, nil, to_gene))
+          table.insert(self.outqueue, deep_copy(best, nil))
         end
       end
     elseif #genes == 1 then

data/speck/companion_html

@@ -0,0 +1,104 @@
+--[[
+  This is an output template for Companion web HTML output.
+]]
+
+function valorzero(val)
+  if val > 0 then
+    return val
+  else
+    return ''
+  end
+end
+
+-- XXX improve performance
+function make_popover_content(aspect_details)
+  local rval = ""
+  for n, node in ipairs(aspect_details.nodes) do
+    rval = rval .. ' ' .. node.ID
+  end
+  return rval
+end
+
+function print_aspects(ftype, aspects)
+  local str = ""
+  local first = true
+  local aspcnt = 0
+  local i = 0
+  for aspect, aspect_details in pairs(aspects) do
+    aspcnt = aspcnt + 1
+  end
+  for aspect, aspect_details in pairs(aspects) do
+    if aspect_details.failures > 0 or  aspect_details.runtime_errors > 0 then
+      str = "fail"
+    else
+      str = "ok"
+    end
+
+    template_print('<tr>')
+    if first then
+      template_print('<td rowspan=' .. aspcnt .. '><i>' .. ftype .. '</i></td>')
+      first = false
+    end
+
+    template_print('<td>' .. aspect .. '</td><td class="text-success">')
+    template_print(valorzero(aspect_details.successes))
+    template_print('</td><td>')
+    if aspect_details.failures > 0 then
+      template_print('<a class="label label-danger" data-toggle="popover" title="Failed items" data-placement="left" data-content="' .. make_popover_content(aspect_details)  .. '">')
+    end
+    template_print(valorzero(aspect_details.failures))
+    if aspect_details.failures > 0 then
+      template_print('</a>')
+    end
+    template_print('</td><td>')
+    if aspect_details.runtime_errors > 0 then
+      template_print('<a class="label label-danger" data-toggle="popover" title="Errored items" data-placement="left" data-content="' .. make_popover_content(aspect_details)  .. '">')
+    end
+    template_print(valorzero(aspect_details.runtime_errors))
+    if aspect_details.runtime_errors > 0 then
+      template_print('</a>')
+    end
+    template_print('</td></tr>')
+    i = i + 1
+  end
+end
+
+template_print([[<i class="glyphicon glyphicon-ok" aria-hidden="true"></i> = confirmed aspects &emsp; <i class="glyphicon glyphicon-warning-sign" aria-hidden="true"></i> = aspect violations &emsp; <i class="glyphicon glyphicon-fire" aria-hidden="true"></i> = runtime errors while checking]])
+
+template_print("<h3>Features</h3>")
+template_print([[<table class="table table-striped table-condensed" style="border-collapse:collapse;">
+    <thead>
+        <tr>
+            <th>type</td>
+            <th>aspect</th>
+            <th><i class="glyphicon glyphicon-ok" aria-hidden="true"></i></th>
+            <th><i class="glyphicon glyphicon-warning-sign" aria-hidden="true"></i></th>
+            <th><i class="glyphicon glyphicon-fire" aria-hidden="true"></i></th>
+        </tr>
+    </thead>
+    <tbody>]])
+for feature, aspects in pairs(features) do
+  local str
+  for aspect, aspect_details in pairs(aspects) do
+    if aspect_details.failures > 0 or  aspect_details.runtime_errors > 0 then
+      str = "fail"
+      break
+    else
+      str = "ok"
+    end
+  end
+  print_aspects(feature, aspects)
+end
+template_print("</tbody></table>\n")
+
+if #warnings > 0 then
+  template_print("<h3>Warnings</h3>")
+  template_print("<ul>\n")
+  for _, msg in ipairs(warnings) do
+    template_print("<li>" .. msg .. "</li>\n")
+  end
+  template_print("</ul>\n")
+  template_print("</div></li>\n")
+
+  template_print("</ul>\n")
+end

data/speck/output_check.lua

@@ -33,23 +33,36 @@ end
       ====  ]]
 
 describe.feature("gene", function(gene)
+  local valid_child_types = {"mRNA","tRNA","rRNA","snoRNA","snRNA", "scRNA",
+                             "lncRNA", "ncRNA", "promoter"}
 
   does_not_cross_a_contig_boundary(gene)
 
   it("has only allowed child types", function()
     for f in gene:direct_children() do
-      expect({"mRNA","tRNA","rRNA","snoRNA","snRNA", "scRNA",
-              "lncRNA", "ncRNA", "promoter"}).should_contain(f:get_type())
+      expect(valid_child_types).should_contain(f:get_type())
     end
   end)
 
+  it("has at least one valid child", function()
+    local has_child = false
+    for f in gene:direct_children() do
+      if not has_child and table.contains(valid_child_types, f:get_type()) then
+        has_child = true
+        break
+      end
+    end
+    expect(has_child).should_be(true)
+  end)
+
   it("only contains allowed attributes", function()
     for k,_ in gene:attribute_pairs() do
       expect({"ID", "Name", "comment", "Note", "End_range",
               "Start_range", "Dbxref", "controlled_curation",
               "previous_systematic_id", "fiveEndPartial", "threeEndPartial",
               "literature", "synonym", "eupathdb_uc",
-              "internalGap", "ratt_ortholog"}).should_contain(k)
+              "internalGap", "ratt_ortholog", "original_prot_length",
+              "Target", "has_internal_stop", "has_frameshift"}).should_contain(k)
     end
   end)
 

params_default.config

@@ -43,6 +43,9 @@ params {
     // Weight description file for kinetoplastids
     WEIGHT_FILE = "${baseDir}/data/weight/weight_kinetoplastid.lua"
 
+    // Template for spec check output
+    SPECK_TEMPLATE = "html"
+
     // Transcript (e.g. RNA-seq) evidence file to use for hints generation
     // (GTF as produced by cufflinks/cuffmerge/...)
     //TRANSCRIPT_FILE = ""