Hotfix for some correct stats and parsing model info from bam heade…

…rs (#126)

* Fixing barcode corrected stats in the case of padded barcodes.
* Fixed header parsing for model information.

src/longbow/correct/command.py

@@ -486,7 +486,14 @@ def _correct_barcode_fn(in_queue, out_queue, bam_header_dict, barcode_tag, corre
             if new_bc is not None:
                 read.set_tag(corrected_tag, new_bc)
                 read.set_tag(longbow.utils.constants.COULD_CORRECT_BARCODE_TAG, True)
-                read.set_tag(longbow.utils.constants.BARCODE_CORRECTION_PERFORMED, new_bc != old_bc)
+
+                # Need to do some basic math for padded barcode strings:
+                if len(old_bc) > barcode_length:
+                    end_pad_bases = int((len(old_bc) - barcode_length)/2)
+                    correction_performed = new_bc != old_bc[end_pad_bases:end_pad_bases+barcode_length]
+                else:
+                    correction_performed = new_bc != old_bc
+                read.set_tag(longbow.utils.constants.BARCODE_CORRECTION_PERFORMED, correction_performed)
 
                 read.set_tag(longbow.utils.constants.READ_ADJUSTED_BARCODE_START, offset)
 

src/longbow/utils/bam_utils.py

@@ -430,10 +430,13 @@ def get_model_from_bam_header(header):
 def get_models_from_bam_header(header):
     model_jsons = []
     for pg in header.as_dict()['PG']:
-        if pg['PN'] == 'longbow' and 'annotate' in pg['ID']:
-            desc, models_str= pg['DS'].split('MODEL(s): ')
-            model_json = json.loads(models_str)
-            model_jsons.append(model_json)
+        try:
+            if pg['PN'] == 'longbow' and (pg['ID'].startswith('longbow-annotate') or pg['ID'].startswith('longbow-pad')):
+                desc, models_str = pg['DS'].split('MODEL(s): ')
+                model_json = json.loads(models_str)
+                model_jsons.append(model_json)
+        except KeyError:
+            continue
 
     return model_jsons