Skip to content

Commit

Permalink
Enforce max_clusters is not set with VAEAAE model
Browse files Browse the repository at this point in the history
When using the VAEAAE ensemble model, we need to cluster the full set of seqs,
since the mandatory following deduplication step is meaningless with multiple
clusterings of different subsets of the full sequences.
  • Loading branch information
jakobnissen authored and Paupiera committed Nov 14, 2023
1 parent 3d220b3 commit ff5eac6
Showing 1 changed file with 11 additions and 0 deletions.
11 changes: 11 additions & 0 deletions vamb/__main__.py
Original file line number Diff line number Diff line change
Expand Up @@ -1613,6 +1613,17 @@ def __init__(self, args):
class VAEAAEArguments(BinnerArguments):
def __init__(self, args):
super(VAEAAEArguments, self).__init__(args)

# For the VAE-AAE workflow, we must cluster the full set of sequences,
# else the cluster dereplication step makes no sense, as the different
# clusterings will contain different sets of clusters.
# So, enforce this here
if self.cluster_options.max_clusters is not None:
raise ValueError(
"When using the VAE-AAE model, `max_clusters` (option `-c`) "
"must not be set."
)

self.aae_options = AAEOptions(
nhiddens=args.nhiddens_aae,
nlatent_z=args.nlatent_aae_z,
Expand Down

0 comments on commit ff5eac6

Please sign in to comment.