Now compile model compiles from all kmers found in target for which we train models

inference/compile_model.py

@@ -126,8 +126,12 @@ def main(args):
                     else:
                         warnings.warn(f'model generation failed for {km}, see {args.out_dir}logs. Continuing compilation without it.')
                         kmers.remove(km)
-        else:  # filter out k-mers for which no stored model exists
-            kmers = get_kmer_candidates_16S(kmer_candidates_dict, args.min_nb_models, 0.0001, filter_list=available_mod_kmers)
+        else:  # Return all k-mers
+            kmers = set()
+            for _, kmers_in_seq in kmer_candidates_dict.items():
+                # Prevent duplicate k-mers
+                kmers.update(kmers_in_seq)
+            # kmers = get_kmer_candidates_16S(kmer_candidates_dict, args.min_nb_models, 0.0001, filter_list=available_mod_kmers)
             if not len(kmers):
                 raise ValueError('Sequences do not contain any of available models!')
 
@@ -135,7 +139,9 @@ def main(args):
         compile_list = []
         for kmer in kmers:
             if kmer not in available_mod_kmers:
-                raise ValueError(f'Cannot compile model: no sub-model available for {kmer}')
+                print(f'No model for {kmer}')
+                continue
+                # raise ValueError(f'Cannot compile model: no sub-model available for {kmer}')
             copyfile(f'{available_mod_kmers[kmer]}/nn.h5', f'{tdo}/{kmer}.h5')
             compile_list.append(available_mod_kmers[kmer])