Merge branch 'model-msg' into 'dev'

Tweak some model choosing exception messages

See merge request research/medaka!595

Author: cjw85

medaka/medaka.py

@@ -34,7 +34,7 @@ def __call__(self, parser, namespace, values, option_string=None):
         try:
             model_fp = medaka.models.resolve_model(val)
         except Exception as e:
-            msg = "Error validating model from '--{}' argument: {}."
+            msg = "Error validating model from '--{}' argument: {}"
             raise RuntimeError(msg.format(self.dest, str(e)))
         setattr(namespace, f"{self.dest}_was_given", True)
         setattr(namespace, self.dest, model_fp)
@@ -44,8 +44,9 @@ class AutoModel(argparse.Action):
     def __call__(self, parser, namespace, values, option_string=None):
         variant, input_file = values
         if variant not in {'consensus', 'variant', 'consensus_bacteria'}:
-            raise ValueError("'TYPE' must be one of 'consensus', 'variant'," 
-                             "or 'consensus_bacteria'.")
+            raise ValueError(
+                "'TYPE' must be one of 'consensus', 'variant',"
+                "or 'consensus_bacteria'.")
         bacteria = 'bacteria' in variant
         variant = variant == 'variant'
         model = medaka.models.model_from_basecaller(
@@ -197,11 +198,11 @@ def _model_arg():
         formatter_class=argparse.ArgumentDefaultsHelpFormatter, add_help=False)
     grp = parser.add_mutually_exclusive_group()
     grp.add_argument('--model', action=ResolveModel,
-            default=medaka.options.default_models['consensus'],
-            help="Model to use. Can be a medaka model name or a basecaller model name suffixed with ':consensus' or ':variant'. For example '[email protected]:variant'.")
+        default=medaka.options.default_models['consensus'],
+        help="Model to use. Can be a medaka model name or a basecaller model name suffixed with ':consensus' or ':variant'. For example '[email protected]:variant'.")
     grp.add_argument('--auto_model', nargs=2, action=AutoModel,
-            metavar=("TYPE", "INPUT"), dest='model',
-            help="Automatically choose model according to INPUT. TYPE should be one of 'consensus' or 'variant'.")
+        metavar=("TYPE", "INPUT"), dest='model',
+        help="Automatically choose model according to INPUT. TYPE should be one of 'consensus' or 'variant'.")
     return parser
 
 

medaka/models.py

@@ -52,8 +52,22 @@ def resolve_model(model):
             raise ex
         except Exception:
             logger.warning(err_msg)
+
+        extra_msg = ""
+        if "fast" in model:
+            extra_msg = (
+                " 'fast' basecalling models are not supported by medaka, "
+                "if you have used a 'fast' model to perform basecalling you "
+                "will need to first perform basecalling with a "
+                "'high accuracy' or 'super accuracy' model before using "
+                "medaka.")
         raise ValueError(
-            f"Model {model} is not a known model or existant file.")
+            f"The model '{model}' is not a recognised basecaller model or "
+            "existant file. This could indicate a malformed input file (for "
+            "which medaka was unable to identify correctly the basecaller "
+            "meta-information) or simply be that the model is not supported "
+            f"by medaka.{extra_msg}")
+
     else:
         # check for model in model stores
         for suffix in model_suffixes:
@@ -168,7 +182,7 @@ def model_from_basecaller(fname, variant=False, bacteria=False):
     # replace with bacterial consensus model if needed
     if bacteria and not variant:
         if model in medaka.options.bact_methyl_compatible_models:
-            model = "r1041_e82_400bps_bacterial_methylation"
+            model = medaka.options.bact_methyl_model
         else:
             logger.warning(
                 "WARNING: --bacteria specified but input data was not "
@@ -210,7 +224,13 @@ def _model_from_fastq(fname):
                         "basecall_model_version_id=")[1].split()[0]
                     models.add(model)
                 except Exception:
-                    pass
+                    try:
+                        # rogue non-spec conforming minknow versions
+                        model = rec.comment.split(
+                            "model_version_id=")[1].split()[0]
+                        models.add(model)
+                    except Exception:
+                        pass
     if len(models) > 1:
         # filter out any models without an `@`. These are likely FPs of
         # the search above (there are unversioned models whose name

medaka/options.py

@@ -165,6 +165,7 @@
     'r1041_e82_400bps_bacterial_methylation',
 ]
 
+bact_methyl_model = 'r1041_e82_400bps_bacterial_methylation'
 bact_methyl_compatible_models = [
     'r1041_e82_400bps_hac_v4.2.0', 'r1041_e82_400bps_sup_v4.2.0',
     'r1041_e82_400bps_hac_v4.3.0', 'r1041_e82_400bps_sup_v4.3.0',