remove (now-)unused glfo seq argument

bin/example-output-processing.py

@@ -17,7 +17,7 @@
     for line in reader:
         utils.process_input_line(line)
         utils.add_implicit_info(glfo, line)
-        utils.print_reco_event(glfo['seqs'], line)
+        utils.print_reco_event(line)
         cdr3_bounds = (line['codon_positions']['v'], line['codon_positions']['j'] + 3)
         print ''
         print '  should match the above:'

python/alleleremover.py

@@ -196,7 +196,7 @@ def finalize(self, pcounter, swfo, debug=False):
                 # unpadded_line['seqs'][0] = unpadded_line['seqs'][0][unpadded_line['padlefts'][0] : ]
                 # if unpadded_line['padrights'][0] > 0:
                 #     unpadded_line['seqs'][0] = unpadded_line['seqs'][0][ : -unpadded_line['padrights'][0]]
-                # utils.print_reco_event(self.glfo['seqs'], unpadded_line)
+                # utils.print_reco_event(unpadded_line)
 
         print '    keeping %d %s genes' % (len(self.genes_to_keep), self.region)
         print '    removing %d %s genes: %d with no matches, %d with unconvincing matches (%d / %d queries had their best match removed)' % (len(self.genes_to_remove), self.region, len(set(self.glfo['seqs'][self.region]) - set(easycounts)), len(set(easycounts) - self.genes_to_keep), n_queries_with_removed_genes, len(swfo['queries']))

python/event.py

@@ -148,7 +148,7 @@ def print_event(self):
         line['cdr3_length'] = self.cdr3_length
         line['codon_positions'] = copy.deepcopy(self.final_codon_positions)
         utils.add_implicit_info(self.glfo, line)
-        utils.print_reco_event(self.glfo['seqs'], line, extra_str='    ')
+        utils.print_reco_event(line, extra_str='    ')
 
     # ----------------------------------------------------------------------------------------
     def print_gene_choice(self):

python/partitiondriver.py

@@ -293,7 +293,7 @@ def view_existing_annotations(self):
                     print '   %s' % ':'.join(line['unique_ids'])
                 else:
                     print ''
-                utils.print_reco_event(self.glfo['seqs'], line)
+                utils.print_reco_event(line)
         if len(failed_queries) > 0:
             print '%d failed queries' % len(failed_queries)
 
@@ -1311,7 +1311,7 @@ def process_dummy_d_hack(self, line, debug=False):
         if debug:
             print ''
             print '  dummy d hack for %s' % ' '.join(line['unique_ids'])
-            utils.print_reco_event(self.glfo['seqs'], tmpline, extra_str='    ', label='before')
+            utils.print_reco_event(tmpline, extra_str='    ', label='before')
 
         gl_v_base = None
         if tmpline['v_3p_del'] > 0:
@@ -1361,7 +1361,7 @@ def process_dummy_d_hack(self, line, debug=False):
         after_line = copy.deepcopy(line)
         utils.add_implicit_info(self.glfo, after_line)
         if debug:
-            utils.print_reco_event(self.glfo['seqs'], after_line, extra_str='    ', label='after')
+            utils.print_reco_event(after_line, extra_str='    ', label='after')
 
     # ----------------------------------------------------------------------------------------
     def read_annotation_output(self, annotation_fname, outfname=None, count_parameters=False, parameter_out_dir=None, print_annotations=False):
@@ -1402,7 +1402,7 @@ def read_annotation_output(self, annotation_fname, outfname=None, count_paramete
                     n_invalid_events += 1
                     if self.args.debug:
                         print '      %s padded line invalid' % uidstr
-                        utils.print_reco_event(self.glfo['seqs'], padded_line, extra_str='    ', label='invalid:')
+                        utils.print_reco_event(padded_line, extra_str='    ', label='invalid:')
                     continue
 
                 assert uidstr not in padded_annotations
@@ -1520,7 +1520,7 @@ def print_hmm_output(self, line, print_true=False):
         label = 'inferred:'
         if self.args.seed_unique_id is not None and self.args.seed_unique_id in line['unique_ids']:
             label += '   (found %d sequences clonal to seed %s)' % (len(line['unique_ids']), self.args.seed_unique_id)
-        utils.print_reco_event(self.glfo['seqs'], line, extra_str='    ', label=label, seed_uid=self.args.seed_unique_id)
+        utils.print_reco_event(line, extra_str='    ', label=label, seed_uid=self.args.seed_unique_id)
 
     # ----------------------------------------------------------------------------------------
     def write_annotations(self, annotations, outfname):

python/prutils.py

@@ -102,7 +102,7 @@ def ismuted(ch1, ch2):
     return outstrs
 
 # ----------------------------------------------------------------------------------------
-def print_seq_in_reco_event(germlines, original_line, iseq, extra_str='', label='', one_line=False, seed_uid=None, check_line_integrity=False):
+def print_seq_in_reco_event(original_line, iseq, extra_str='', label='', one_line=False, seed_uid=None, check_line_integrity=False):
     """
     Print ascii summary of recombination event and mutation.
     If <one_line>, then skip the germline lines, and only print the final_seq line.
@@ -135,7 +135,7 @@ def print_seq_in_reco_event(germlines, original_line, iseq, extra_str='', label=
                   + line['vd_insertion'] + ' ' * line['lengths']['d'] + line['dj_insertion'] \
                   + ' ' * (line['lengths']['j'] + line['j_3p_del'] + len(line['jf_insertion']))
     germline_d_start = len(line['fv_insertion']) + line['lengths']['v'] + len(line['vd_insertion']) - line['d_5p_del']
-    germline_d_end = germline_d_start + len(germlines['d'][line['d_gene']])
+    germline_d_end = germline_d_start + line['d_5p_del'] + line['lengths']['d'] + line['d_3p_del']
     d_line = ' ' * (germline_d_start + len(delstrs['v_5p'])) \
              + eroded_seqs_dots['d'] \
              + ' ' * (len(line['j_gl_seq']) + len(line['dj_insertion']) - line['d_3p_del'] + line['j_3p_del'] + len(line['jf_insertion']))
@@ -146,16 +146,19 @@ def print_seq_in_reco_event(germlines, original_line, iseq, extra_str='', label=
     # and the query line
     qrseq_line = ' ' * len(delstrs['v_5p']) + line['seqs'][iseq] + ' ' * line['j_3p_del']
 
+    outstrs = [insert_line, d_line, vj_line, qrseq_line]
+
     if gap_insert_point is not None:  # <gap_insert_point> point is only right here as long as there's no colors in these lines... but there usually almost probably always aren't
-        qrseq_line = qrseq_line[:gap_insert_point] + gapstr + qrseq_line[gap_insert_point:]
-        insert_line = insert_line[:gap_insert_point] + gapstr + insert_line[gap_insert_point:]
-        d_line = d_line[:gap_insert_point] + gapstr + d_line[gap_insert_point:]
+        for istr in [0, 1, 3]:  # everybody except the vj line, which already has the modified interior delstrs above
+            outstrs[istr] = outstrs[istr][:gap_insert_point] + gapstr + outstrs[istr][gap_insert_point:]
+
+    if len(set([len(ostr) for ostr in outstrs])) > 1:  # could put this in a bunch of different places, but things're probably most likely to get screwed up either when initally building the four lines, or dealing with the stupid gaps
+        raise Exception('outstrs not all the same length %s' % [len(ostr) for ostr in outstrs])
 
     chain = utils.get_chain(line['v_gene'])
     if chain != 'h':
         assert line['lengths']['d'] == 0 and len(line['vd_insertion']) == 0
 
-    outstrs = [insert_line, d_line, vj_line, qrseq_line]
     colors = [[[] for _ in range(len(ostr))] for ostr in outstrs]
     outstrs, colors = indel_shenanigans(outstrs, colors, line['indelfos'][iseq], iseq)
     outstrs = add_colors(outstrs, colors, line)

python/utils.py

@@ -406,7 +406,7 @@ def convert_from_adaptive_headers(glfo, line, uid=None, only_dj_rearrangements=F
 
     if print_it:
         add_implicit_info(glfo, newline)
-        print_reco_event(glfo['seqs'], newline, label=uid)
+        print_reco_event(newline, label=uid)
 
     # still need to convert to integers/lists/whatnot (?)
 
@@ -996,7 +996,7 @@ def print_true_events(glfo, reco_info, line, print_naive_seqs=False, extra_str='
     true_naive_seqs = []
     for uids in get_true_partition(reco_info, ids=line['unique_ids']):  # make a multi-seq line that has all the seqs from this clonal family
         multiline = synthesize_multi_seq_line(uids, reco_info)
-        print_reco_event(glfo['seqs'], multiline, extra_str=extra_str, label=color('green', 'true:'))
+        print_reco_event(multiline, extra_str=extra_str, label=color('green', 'true:'))
         true_naive_seqs.append(multiline['naive_seq'])
 
     if print_naive_seqs:
@@ -1005,9 +1005,9 @@ def print_true_events(glfo, reco_info, line, print_naive_seqs=False, extra_str='
             color_mutants(tseq, line['naive_seq'], print_result=True, print_hfrac=True, ref_label='true ', extra_str='          ')
 
 # ----------------------------------------------------------------------------------------
-def print_reco_event(germlines, line, one_line=False, extra_str='', label='', seed_uid=None):
+def print_reco_event(line, one_line=False, extra_str='', label='', seed_uid=None):
     for iseq in range(len(line['unique_ids'])):
-        prutils.print_seq_in_reco_event(germlines, line, iseq, extra_str=extra_str, label=(label if iseq==0 else ''), one_line=(iseq>0), seed_uid=seed_uid)
+        prutils.print_seq_in_reco_event(line, iseq, extra_str=extra_str, label=(label if iseq==0 else ''), one_line=(iseq>0), seed_uid=seed_uid)
 
 #----------------------------------------------------------------------------------------
 def sanitize_name(name):

python/waterer.py

@@ -149,7 +149,7 @@ def finalize(self, cachefname=None, just_read_cachefile=False):
             if self.debug and not self.args.is_data and len(self.remaining_queries) > 0:
                 print 'true annotations for remaining events:'
                 for qry in self.remaining_queries:
-                    utils.print_reco_event(self.glfo['seqs'], self.reco_info[qry], extra_str='      ', label='true:')
+                    utils.print_reco_event(self.reco_info[qry], extra_str='      ', label='true:')
 
         self.info['remaining_queries'] = self.remaining_queries
 
@@ -772,8 +772,8 @@ def add_to_info(self, infoline):
             inf_label = '      ' + utils.kbound_str({r : infoline['k_' + r] for r in ['v', 'd']})
             if not self.args.is_data:
                 inf_label = 'inferred: ' + inf_label
-                utils.print_reco_event(self.glfo['seqs'], self.reco_info[qname], extra_str='    ', label=utils.color('green', 'true:'))
-            utils.print_reco_event(self.glfo['seqs'], self.info[qname], extra_str='    ', label=inf_label)
+                utils.print_reco_event(self.reco_info[qname], extra_str='    ', label=utils.color('green', 'true:'))
+            utils.print_reco_event(self.info[qname], extra_str='    ', label=inf_label)
 
         if self.pcounter is not None:
             self.pcounter.increment(self.info[qname])
@@ -1093,7 +1093,7 @@ def remove_framework_insertions(self, debug=False):
                 if swfo['indelfos'][0]['reversed_seq'] != swfo['seqs'][0]:
                     print '%s reversed seq not same as seq:\n%s\n%s' % (utils.color('red', 'warning'), swfo['indelfos'][0]['reversed_seq'], swfo['seqs'][0])
 
-            # utils.print_reco_event(self.glfo['seqs'], swfo)
+            # utils.print_reco_event(swfo)
             utils.remove_all_implicit_info(swfo)
             fv_len = len(swfo['fv_insertion'])
             jf_len = len(swfo['jf_insertion'])
@@ -1109,7 +1109,7 @@ def remove_framework_insertions(self, debug=False):
             swfo['jf_insertion'] = ''
 
             utils.add_implicit_info(self.glfo, swfo)
-            # utils.print_reco_event(self.glfo['seqs'], swfo)
+            # utils.print_reco_event(swfo)
 
             # *sigh* not super happy about it, but I think the best way to handle this is to also remove these bases from the simulation info
             if self.reco_info is not None: