Merge branch 'remora_motif_expand' into 'master'

Remora motif expand See merge request algorithm/megalodon!88
nanoporetech · Jan 26, 2022 · 16f9df2 · 16f9df2
2 parents a693e9a + caed143
commit 16f9df2
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 11 deletions.
diff --git a/megalodon/backends.py b/megalodon/backends.py
@@ -1107,19 +1107,23 @@ def pyguppy_set_model_attributes(
             )
         self.is_flipflop = init_called_read.model_type == FF_GUPPY_NAME
         self.is_crf = init_called_read.model_type == CRF_GUPPY_NAME
-        if self.is_crf:
-            LOGGER.info("CRF models are not fully supported.")
 
         self.stride = init_called_read.model_stride
         if remora_model_filename is not None or remora_model_spec is not None:
             self.is_remora_mod = True
             from remora import model_util
 
             if remora_model_filename is not None:
+                LOGGER.debug(
+                    f"Loading Remora model from file: {remora_model_filename}"
+                )
                 _, remora_metadata = model_util.load_model(
                     remora_model_filename, quiet=True
                 )
             else:
+                LOGGER.debug(
+                    f"Loading Remora model from specs: {remora_model_spec}"
+                )
                 _, remora_metadata = model_util.load_model(
                     pore=remora_model_spec[0],
                     basecall_model_type=remora_model_spec[1],
@@ -1130,8 +1134,7 @@ def pyguppy_set_model_attributes(
                     quiet=True,
                 )
             self.ordered_mod_long_names = remora_metadata["mod_long_names"]
-            motif, motif_offset = remora_metadata["motif"]
-            can_idx = "ACGT".find(motif[motif_offset]) + 1
+            can_idx = "ACGT".find(remora_metadata["can_base"]) + 1
             self.output_alphabet = (
                 "ACGT"[:can_idx]
                 + remora_metadata["mod_bases"]

diff --git a/megalodon/mods.py b/megalodon/mods.py
@@ -2298,12 +2298,11 @@ def __init__(
             self.remora_RemoraRead = data_chunks.RemoraRead
             self.load_remora_model()
             self.mod_bases = self.remora_metadata["mod_bases"]
-            motif, motif_offset = self.remora_metadata["motif"]
-            can_base = motif[motif_offset]
+            can_base = self.remora_metadata["can_base"]
             can_idx = "ACGT".find(can_base) + 1
             if can_idx == 0:
                 raise mh.MegaError(
-                    f"Invalid Remora model motif {motif}:{motif_offset}"
+                    f"Invalid Remora model canonical base {can_base}"
                 )
             self.alphabet = "ACGT"
             self.output_alphabet = (
@@ -2322,12 +2321,12 @@ def __init__(
                 for mod_b, mln in self.mod_long_names
             )
             LOGGER.info(f"Loaded Remora model calls modified bases: {mod_str}")
-            if all_mod_motifs_raw is None:
+            if all_mod_motifs_raw is not None:
                 LOGGER.info(
-                    f'Setting --mod-motif to "{motif} {motif_offset}" loaded '
-                    "from Remora model"
+                    "Modified base motifs derived from Remora model. "
+                    "Ignoring --mod-motif."
                 )
-                all_mod_motifs_raw = [(self.mod_bases, motif, motif_offset)]
+                all_mod_motifs_raw = None
             # only applicable to flip-flop modbases
             self.can_mods_offsets = None
             self.str_to_int_mod_labels = None