Fall back to Gotoh alignment in genome coverage, as part of #589.

Also upgrade to Ubuntu 20.04 in GitHub Actions, because 16.04 is losing support.

.github/workflows/build-and-test.yml

@@ -2,7 +2,7 @@ name: Build and Test MiCall
 on: push
 jobs:
   build:
-    runs-on: ubuntu-16.04
+    runs-on: ubuntu-20.04
     steps:
       - uses: actions/checkout@v2
       - name: Set up Python

micall/core/aln2counts.py

@@ -16,13 +16,12 @@
 import csv
 from csv import DictWriter
 from itertools import groupby
-from operator import itemgetter, attrgetter
+from operator import itemgetter
 import os
 from pathlib import Path
 
 import gotoh
 import yaml
-from mappy import Aligner
 
 from micall.core.project_config import ProjectConfig, G2P_SEED_NAME
 from micall.core.remap import PARTIAL_CONTIG_SUFFIX, REVERSED_CONTIG_SUFFIX
@@ -642,12 +641,9 @@ def write_sequence_coverage_counts(self,
             coordinate_name = None
             alignments = []
         else:
-            coordinate_seq = self.projects.getReference(coordinate_name)
-            aligner = Aligner(seq=coordinate_seq, preset='map-ont')
-            alignments = sorted(aligner.map(consensus), key=attrgetter('q_st'))
-            alignments = [alignment
-                          for alignment in alignments
-                          if alignment.is_primary]
+            consensus_aligner = ConsensusAligner(self.projects)
+            consensus_aligner.start_contig(coordinate_name, consensus)
+            alignments = consensus_aligner.alignments
         wire = Wire()
         bead_end = source_end = 0
         for alignment in alignments:

micall/tests/test_aln2counts_report.py

@@ -934,6 +934,7 @@ def test_contig_coverage_report_merged_contigs(sequence_report):
     Contig 3-R1 TTTAGG -> FR
 
     Contig 1 and 3 have been combined into R1 with KFR.
+    R1 is KFR, and R2 is KFGPR.
     """
     # refname,qcut,rank,count,offset,seq
     aligned_reads1 = prepare_reads("1_3-R1-seed,15,0,5,0,AAATTT\n"
@@ -948,39 +949,39 @@ def test_contig_coverage_report_merged_contigs(sequence_report):
 
     expected_text = """\
 contig,coordinates,query_nuc_pos,refseq_nuc_pos,dels,coverage,link
-1_3-R1-seed,R1-seed,1,1,0,5,U
-1_3-R1-seed,R1-seed,2,2,0,5,U
-1_3-R1-seed,R1-seed,3,3,0,5,U
-1_3-R1-seed,R1-seed,4,4,0,7,U
-1_3-R1-seed,R1-seed,5,5,0,7,U
-1_3-R1-seed,R1-seed,6,6,0,7,U
-1_3-R1-seed,R1-seed,7,7,0,2,U
-1_3-R1-seed,R1-seed,8,8,0,2,U
-1_3-R1-seed,R1-seed,9,9,0,2,U
-contig-1-R1-seed,R1-seed,1,1,,,U
-contig-1-R1-seed,R1-seed,2,2,,,U
-contig-1-R1-seed,R1-seed,3,3,,,U
-contig-1-R1-seed,R1-seed,4,4,,,U
-contig-1-R1-seed,R1-seed,5,5,,,U
-contig-1-R1-seed,R1-seed,6,6,,,U
-contig-3-R1-seed,R1-seed,1,1,,,U
-contig-3-R1-seed,R1-seed,2,2,,,U
-contig-3-R1-seed,R1-seed,3,3,,,U
-contig-3-R1-seed,R1-seed,4,4,,,U
-contig-3-R1-seed,R1-seed,5,5,,,U
-contig-3-R1-seed,R1-seed,6,6,,,U
-2-R2-seed,R2-seed,1,1,0,4,U
-2-R2-seed,R2-seed,2,2,0,4,U
-2-R2-seed,R2-seed,3,3,0,4,U
-2-R2-seed,R2-seed,4,4,0,4,U
-2-R2-seed,R2-seed,5,5,0,4,U
-2-R2-seed,R2-seed,6,6,0,4,U
-contig-2-R2-seed,R2-seed,1,1,,,U
-contig-2-R2-seed,R2-seed,2,2,,,U
-contig-2-R2-seed,R2-seed,3,3,,,U
-contig-2-R2-seed,R2-seed,4,4,,,U
-contig-2-R2-seed,R2-seed,5,5,,,U
-contig-2-R2-seed,R2-seed,6,6,,,U
+1_3-R1-seed,R1-seed,1,1,0,5,M
+1_3-R1-seed,R1-seed,2,2,0,5,M
+1_3-R1-seed,R1-seed,3,3,0,5,M
+1_3-R1-seed,R1-seed,4,4,0,7,M
+1_3-R1-seed,R1-seed,5,5,0,7,M
+1_3-R1-seed,R1-seed,6,6,0,7,M
+1_3-R1-seed,R1-seed,7,7,0,2,M
+1_3-R1-seed,R1-seed,8,8,0,2,M
+1_3-R1-seed,R1-seed,9,9,0,2,M
+contig-1-R1-seed,R1-seed,1,1,,,M
+contig-1-R1-seed,R1-seed,2,2,,,M
+contig-1-R1-seed,R1-seed,3,3,,,M
+contig-1-R1-seed,R1-seed,4,4,,,M
+contig-1-R1-seed,R1-seed,5,5,,,M
+contig-1-R1-seed,R1-seed,6,6,,,M
+contig-3-R1-seed,R1-seed,1,4,,,M
+contig-3-R1-seed,R1-seed,2,5,,,M
+contig-3-R1-seed,R1-seed,3,6,,,M
+contig-3-R1-seed,R1-seed,4,7,,,M
+contig-3-R1-seed,R1-seed,5,8,,,M
+contig-3-R1-seed,R1-seed,6,9,,,M
+2-R2-seed,R2-seed,1,7,0,4,M
+2-R2-seed,R2-seed,2,8,0,4,M
+2-R2-seed,R2-seed,3,9,0,4,M
+2-R2-seed,R2-seed,4,10,0,4,M
+2-R2-seed,R2-seed,5,11,0,4,M
+2-R2-seed,R2-seed,6,12,0,4,M
+contig-2-R2-seed,R2-seed,1,7,,,M
+contig-2-R2-seed,R2-seed,2,8,,,M
+contig-2-R2-seed,R2-seed,3,9,,,M
+contig-2-R2-seed,R2-seed,4,10,,,M
+contig-2-R2-seed,R2-seed,5,11,,,M
+contig-2-R2-seed,R2-seed,6,12,,,M
 """
 
     report_file = StringIO()