Fix #64

.gitignore

@@ -49,3 +49,4 @@ hg3*.fa*
 paper/workflow/results
 *.ipynb
 *.tar
+.pytest_cache/

CHANGES.md

@@ -2,12 +2,11 @@
 
 dev
 ---
-* Added four new tools to the benchmarks:
+* Added two new tools to the benchmarks:
   * fastp
   * Cutadapt
-  * Sickle
-  * SeqPrep
 * Updated versions of several tools used in the paper. Rebuilt containers and pushed them to Dockerhub.
+* Fix #64: InsertAligner not respecting match_adapter_wildcards and match_read_wildcards options.
 
 v1.1.17 (2018.01.13)
 --------------------

README.md

@@ -117,6 +117,7 @@ The citation for the original Cutadapt paper is:
     * Create Galaxy tool description using [argparse2tool](https://github.com/erasche/argparse2tool#cwl-specific-functionality).
 * Improve documentation (#24)
 * Port over improvements in latest versions of Cutadapt https://cutadapt.readthedocs.io/en/stable/
+* Switch to using entry point instead of Atropos executable.
 
 ### 1.3
 
@@ -151,6 +152,7 @@ The citation for the original Cutadapt paper is:
 
 ### 1.6
 
+* Switch to using Click for CLI.
 * Implement a public plugin API.
 * Add more logging and convert log messages from old-style to new-style format strings.
 * Add option to estimate bisulfite conversion rate from filled-in cytosine methylation status in reads that were MspI-digested.

atropos/adapters/__init__.py

@@ -367,7 +367,6 @@ def match_to(self, read):
                 self, read)
 
         # try approximate matching
-        alignment = None
         if not self.indels and self.where in (PREFIX, SUFFIX):
             if self.where == PREFIX:
                 alignment = align.compare_prefixes(

atropos/align/__init__.py

@@ -245,7 +245,8 @@ def __init__(
             insert_max_rmp=1E-6, adapter_max_rmp=0.001,
             min_insert_overlap=1, max_insert_mismatch_frac=0.2,
             min_adapter_overlap=1, max_adapter_mismatch_frac=0.2,
-            adapter_check_cutoff=9, base_probs=None):
+            adapter_check_cutoff=9, base_probs=None,
+            adapter_wildcards=True, read_wildcards=False):
         self.adapter1 = adapter1
         self.adapter1_len = len(adapter1)
         self.adapter2 = adapter2
@@ -260,6 +261,8 @@ def __init__(
         self.adapter_check_cutoff = adapter_check_cutoff
         self.base_probs = base_probs or dict(
             match_prob=0.25, mismatch_prob=0.75)
+        self.adapter_wildcards = adapter_wildcards
+        self.read_wildcards = read_wildcards
         self.aligner = MultiAligner(
             max_insert_mismatch_frac,
             START_WITHIN_SEQ1 | STOP_WITHIN_SEQ2,
@@ -299,8 +302,14 @@ def _match(insert_match, offset, insert_match_size, prob): # pylint disable=unus
             # the alignment will fail. We need to use a comparison that is a bit
             # more forgiving.
 
-            a1_match = compare_prefixes(seq1[insert_match_size:], self.adapter1)
-            a2_match = compare_prefixes(seq2[insert_match_size:], self.adapter2)
+            a1_match = compare_prefixes(
+                seq1[insert_match_size:], self.adapter1,
+                wildcard_ref=self.adapter_wildcards,
+                wildcard_query=self.read_wildcards)
+            a2_match = compare_prefixes(
+                seq2[insert_match_size:], self.adapter2,
+                wildcard_ref=self.adapter_wildcards,
+                wildcard_query=self.read_wildcards)
             adapter_len = min(offset, self.adapter1_len, self.adapter2_len)
             max_adapter_mismatches = round(
                 adapter_len * self.max_adapter_mismatch_frac)

atropos/commands/trim/__init__.py

@@ -379,7 +379,9 @@ def __call__(self):
                         max_adapter_mismatch_frac=\
                             options.insert_match_adapter_error_rate,
                         match_probability=match_probability,
-                        insert_max_rmp=options.insert_max_rmp)
+                        insert_max_rmp=options.insert_max_rmp,
+                        read_wildcards=options.match_read_wildcards,
+                        adapter_wildcards=options.match_adapter_wildcards)
                 else:
                     a1_args = dict(
                         adapters=adapters1,

atropos/commands/trim/modifiers.py

@@ -381,7 +381,8 @@ def __init__(
         self.adapter2 = adapter2
         self.aligner = InsertAligner(
             adapter1.sequence, adapter2.sequence,
-            min_insert_overlap=min_insert_overlap, **aligner_args)
+            min_insert_overlap=min_insert_overlap,
+            **aligner_args)
         self.min_insert_len = min_insert_overlap
         self.action = action
         self.symmetric = symmetric