protein_summary.rb in mspire-0.2.4

- old
+ new

@@ -4,10 +4,11 @@
 require 'hash_by'
 require 'optparse'
 require 'ostruct'
 require 'spec_id'
 require 'spec_id/precision'
+require 'gi'
 
 #############################################################
 # GLOBALS:
 PRECISION_PROGRAM_BASE = 'precision'
 DEF_PREFIX = "INV_"
@@ -208,17 +209,30 @@
     ####    #readable_previous_fppr_rate_percent = sprintf("%.2f", previous_fppr_rate_percent)
 
   # returns a string of the table rows
   # false_positive_rate (give as a %) is the cutoff mark
   # returns the number of proteins at the desired_fppr (if given)
-  def table_rows(uniq_prots, prefix, false_positive_rate_percent, num_cols, desired_fppr, actual_percent_fp, peptide_count_filename=nil)
+  def table_rows(uniq_prots, prefix, false_positive_rate_percent, num_cols, desired_fppr, actual_percent_fp, annotations=nil, peptide_count_filename=nil)
     prot_cnt = 0
+    an_cnt = 0
+ 
     uniq_prots.map do |prot| 
       tr do
         prot_cnt += 1
         gi = accession(prot._protein_name)
-        tds([prot_cnt, prot._probability, ref_html(gi, prot._protein_name), prot.annotation.first._protein_description, prot._percent_coverage, peptide_cell(prot_cnt, prot._unique_stripped_peptides.split('+')), prot._total_number_peptides, prot._pct_spectrum_ids])
+
+        if annotations
+          protein_description = annotations[an_cnt]
+          an_cnt += 1
+        else
+          if prot.annotation.size > 0
+            protein_description = prot.annotation.first._protein_description
+          else
+            protein_description = 'NA'
+          end
+        end
+        tds([prot_cnt, prot._probability, ref_html(gi, prot._protein_name), protein_description, prot._percent_coverage, peptide_cell(prot_cnt, prot._unique_stripped_peptides.split('+')), prot._total_number_peptides, prot._pct_spectrum_ids])
       end
     end.join
   end
 
   def print_html_pieces(file, *pieces)
@@ -308,12 +322,19 @@
       filtered_sorted_prots = filtered_sorted_prots[0,num_prots]
     end
 
     output_peptide_counts_file(filtered_sorted_prots, opt.peptide_count) if opt.peptide_count
 
+    # get an array of annotations (or nil if no option)
+    annotations =
+      if opt.get_annotation
+        gis = filtered_sorted_prots.map {|prot| accession(prot._protein_name) }
+        GI.gi2annot(gis) 
+      end
+
     table_string = table do 
-      tr{theaders} + table_rows(filtered_sorted_prots, opt.f, actual_percent_fp, num_cols, opt.c.to_f, actual_percent_fp, opt.peptide_count)
+      tr{theaders} + table_rows(filtered_sorted_prots, opt.f, actual_percent_fp, num_cols, opt.c.to_f, actual_percent_fp, annotations, opt.peptide_count)
     end
     er_info = opt.precision ? error_info(file) : ""
     html_pieces = [outfn, header, fppr_output_as_html, er_info, file_info(file), protproph_script_info, num_prots_html, table_string, trailer]
     print_html_pieces(*html_pieces)
   end # proph_output
@@ -408,9 +429,11 @@
       op.separator("                                     type '#{PRECISION_PROGRAM_BASE}.rb' for details")
       op.separator ""
       op.separator "Specific to ProteinProphet (with no concatenated DB):"
       op.on("-c", "--cutoff percent", "false positive predictive rate (FPPR)% for given cutoff") {|v| opt.c = v }
       op.on("--cut_at percent", "only reports proteins within FPPR %") {|v| opt.cut_at = v }
+      op.on("--get_annotation", "retrieves annotation by gi code") {|v| opt.get_annotation = v}
+      op.separator "             (use if your proteins have gi's but no annotation) "
     end
 
     opts.parse!(argv)
 
     if argv.size < 1