def create_partial_pickle(flnc_files, chunked_nfl_files, out_pickle):
"""
Parameters:
flnc_files -- full-length non-chimeric files in bins
chunked_nfl_files -- chunked non-chimeric files
out_pickle -- output pickle for saving PolishChunkTask objects
"""
n_bins = len(flnc_files)
n_nfl_chunks = max(1, len(chunked_nfl_files))
log.info("Writing %s ice_partial chunk tasks to %s.",
str(n_bins * n_nfl_chunks), out_pickle)
p = ChunkTasksPickle()
for i, flnc_file in enumerate(flnc_files):
log.debug("Processing cluster bin index=%s.", i)
cluster_out_dir = _get_cluster_out_dir(flnc_file)
for j, nfl_file in enumerate(chunked_nfl_files):
# Create Partial chunk tasks.
task_ = PartialChunkTask(cluster_bin_index=i,
flnc_file=flnc_file,
cluster_out_dir=cluster_out_dir,
nfl_file=nfl_file,
nfl_index=j,
n_nfl_chunks=n_nfl_chunks)
p.append(task_)
p.write(out_pickle)
log.info("Saved %s partial chunk tasks to %s.", str(n_bins * n_nfl_chunks),
out_pickle)
def create_polish_pickle(n_polish_chunks_in_bins, flnc_files, out_pickle):
"""
Parameters:
n_polish_chunks_in_bins -- number of ice_polish chunks in each bin
flnc_files -- full-length non-chimeric files in bins
out_pickle -- output pickle for saving PolishChunkTask objects
"""
n_bins = len(flnc_files)
assert isinstance(n_polish_chunks_in_bins, list)
assert len(n_polish_chunks_in_bins) == n_bins
log.info("Writing %s ice_polish chunk tasks to %s.",
str(sum(n_polish_chunks_in_bins)), out_pickle)
p = ChunkTasksPickle()
for i, flnc_file in enumerate(flnc_files):
log.debug("Creating %s ice_polish chunks for bin index=%s.",
str(n_polish_chunks_in_bins[i]), str(i))
cluster_out_dir = _get_cluster_out_dir(flnc_file)
for j in range(0, n_polish_chunks_in_bins[i]):
# Create Polish chunk tasks.
task_ = PolishChunkTask(cluster_bin_index=i,
flnc_file=flnc_file,
cluster_out_dir=cluster_out_dir,
polish_index=j,
n_polish_chunks=n_polish_chunks_in_bins[i])
p.append(task_)
p.write(out_pickle)
log.info("Saved %s polish chunk tasks to %s.",
str(sum(n_polish_chunks_in_bins)), out_pickle)
def create_cluster_pickle(flnc_files, out_pickle):
"""Create cluster chunk task pickle.
Parameters:
n_bins -- number of bins
flnc_files -- full-length non-chimeric files in bins
out_pickle -- output pickle for saving ClusterChunkTask objects
"""
n_bins = len(flnc_files)
log.info("Writing %s cluster chunk tasks to %s.", str(n_bins), out_pickle)
p = ChunkTasksPickle()
for i, flnc_file in enumerate(flnc_files):
log.debug("Processing cluster bin index=%s.", i)
cluster_out_dir = _get_cluster_out_dir(flnc_file)
# Create Cluster chunk tasks.
task_ = ClusterChunkTask(cluster_bin_index=i, flnc_file=flnc_file,
cluster_out_dir=cluster_out_dir)
p.append(task_)
p.write(out_pickle)
log.info("Saved %s cluster chunk tasks to %s.", str(n_bins), out_pickle)