Fixed seek/tell bug for dysgu-call

dysgu/call_component.pyx

@@ -930,7 +930,7 @@ cdef tuple break_ops(positions, precise, int limit, float median_pos):
             ops.append((i + v + 1, -1, 1))
         else:
             ops.append((i + v, -1, 1))
-    ops = sorted(ops, reverse=limit < 0)
+    ops.sort(reverse=limit < 0)
     cdef int cum_sum = 0
     cdef int max_sum = 0
     cdef int max_sum_i = 0
@@ -1386,10 +1386,10 @@ cdef call_from_reads(u_reads_info, v_reads_info, int insert_size, int insert_std
             svtypes_counts[2].append(i)
         elif i.svtype == "TRA":
             svtypes_counts[3].append(i)
-    svtypes_res = sorted(svtypes_counts, key=sort_by_length, reverse=True)
+    svtypes_counts.sort(key=sort_by_length, reverse=True)
     results = []
     cdef EventResult_t er
-    for sub_informative in svtypes_res:
+    for sub_informative in svtypes_counts:
         if len(sub_informative) >= min_support:
             svtype = sub_informative[0].svtype
             if svtype == "INV" or svtype == "TRA":
@@ -1617,7 +1617,8 @@ cdef get_reads(infile, nodes_info, buffered_reads, n2n, bint add_to_buffer, site
             aligns.append((n, buffered_reads[int_node]))
             continue
         fpos.append((n, int_node))
-    for node, int_node in sorted(fpos, key=fpos_srt):
+    fpos.sort(key=fpos_srt)
+    for node, int_node in fpos:
         infile.seek(node.tell)
         try:
             a = next(infile)
@@ -1643,7 +1644,7 @@ cdef get_reads(infile, nodes_info, buffered_reads, n2n, bint add_to_buffer, site
                     if add_to_buffer:
                         buffered_reads[int_node] = a
                     break
-    #         else:
+            # else:
     #             if has_index:
     #                 nn = n2n[int_node]
     #                 hash_names[(nn.hash_name, nn.flag, nn.pos, nn.chrom)] = nn

dysgu/cluster.pyx

@@ -395,8 +395,8 @@ def merge_events(potential, max_dist, tree, paired_end=False, try_rev=False, pic
     node_to_event = {i.event_id: i for i in potential}
     cdef int k
     for grp in components:
-        c = [node_to_event[n] for n in grp]
-        best = sorted(c, key=srt_func, reverse=True)
+        best = [node_to_event[n] for n in grp]
+        best.sort(key=srt_func, reverse=True)
         w0 = best[0]
         if not pick_best:
             weight = w0.pe + w0.supp + w0.spanning

dysgu/coverage.pyx

@@ -161,10 +161,10 @@ cdef class GenomeScanner:
                 else:
                     f_iter = self.input_bam.fetch(until_eof=True)
                 for aln in f_iter:
-                    # if aln.flag & 1284 or aln.mapq < mq_thresh or aln.cigartuples is None:
-                    #     continue
+
                     cigar_l = aln._delegate.core.n_cigar
                     if cigar_l == 0 or aln.flag & 1284:
+                        tell = 0 if self.no_tell else self.input_bam.tell()
                         continue
 
                     if aln.rname != self.current_tid:
@@ -182,7 +182,6 @@ cdef class GenomeScanner:
                         good_read = True
                     elif not aln.flag & 2:
                         good_read = True
-                    # cigar_l = aln._delegate.core.n_cigar
                     cigar_p = bam_get_cigar(aln._delegate)
                     for i in range(cigar_l):
                         cigar_value = cigar_p[i]
@@ -199,12 +198,11 @@ cdef class GenomeScanner:
                         elif opp == 0 or opp == 7 or opp == 8:
                             self.cpp_cov_track.add(pos + index_start, pos + index_start + length)
                             index_start += length
-                    if aln.mapq < mq_thresh:
-                        continue
-                    if not good_read:
-                        continue
-                    if not self.cpp_cov_track.cov_val_good(self.current_tid, aln.rname, pos):
+
+                    if aln.mapq < mq_thresh or not good_read or not self.cpp_cov_track.cov_val_good(self.current_tid, aln.rname, pos):
+                        tell = 0 if self.no_tell else self.input_bam.tell()
                         continue
+
                     self._add_to_bin_buffer(aln, tell)
                     tell = 0 if self.no_tell else self.input_bam.tell()
                     while len(self.staged_reads) > 0:
@@ -460,9 +458,9 @@ cdef class GenomeScanner:
         cdef float current_coverage
         if self.current_chrom != rname:
             self.current_chrom = rname
-        in_roi = False
-        if self.overlap_regions:
-            in_roi = intersecter(self.overlap_regions, a.rname, apos, apos+1)
+        # in_roi = False
+        # if self.overlap_regions:
+        #     in_roi = intersecter(self.overlap_regions, a.rname, apos, apos+1)
         if rname == self.current_chrom and bin_pos == self.current_pos:
             self.current_bin.append((a, tell))
         else:

dysgu/graph.pyx

@@ -704,7 +704,7 @@ class AlignmentsSA:
                 query_end = start_temp + query_end - query_start
                 query_start = start_temp
             query_aligns.append(AlnBlock(query_start, query_end, ref_start, ref_end, gettid(sa[0]), int(sa[4]), sa[2], False))
-        query_aligns = sorted(query_aligns)
+        query_aligns.sort()
         cdef int index = 0
         for i, item in enumerate(query_aligns):
             if item.this: