class Clusterer:
def __init__(self):
self.clust = Deduplicator()
self.seqio = SequenceIO()
self.seq_library = {}
self.orfm_regex = OrfM.regular_expression()
def uncluster_annotations(self, input_annotations, reverse_pipe):
'''
Update the annotations hash provided by pplacer to include all
representatives within each cluster
Parameters
----------
input_annotations : hash
Classifications for each representative sequence of the clusters.
each key being the sequence name, and the entry being the taxonomy
string as a list.
reverse_pipe : bool
True/False, whether the reverse reads pipeline is being followed.
Returns
-------
output_annotations : hash
An updated version of the above, which includes all reads from
each cluster
'''
output_annotations = {}
for placed_alignment_file_path, clusters in self.seq_library.items():
if reverse_pipe and placed_alignment_file_path.endswith(
"_reverse_clustered.fa"):
continue
placed_alignment_file = os.path.basename(
placed_alignment_file_path)
cluster_classifications = input_annotations[placed_alignment_file]
if reverse_pipe:
placed_alignment_base = placed_alignment_file.replace(
'_forward_clustered.fa', '')
else:
placed_alignment_base = placed_alignment_file.replace(
'_clustered.fa', '')
output_annotations[placed_alignment_base] = {}
for rep_read_name, rep_read_taxonomy in cluster_classifications.items(
):
if reverse_pipe:
orfm_regex = OrfM.regular_expression()
clusters = {(orfm_regex.match(key).groups(0)[0]
if orfm_regex.match(key) else key): item
for key, item in iter(clusters.items())}
for read in clusters[rep_read_name]:
output_annotations[placed_alignment_base][
read.name] = rep_read_taxonomy
return output_annotations
def cluster(self, input_fasta_list, reverse_pipe):
'''
cluster - Clusters reads at 100% identity level and writes them to
file. Resets the input_fasta variable as the FASTA file containing the
clusters.
Parameters
----------
input_fasta_list : list
list of strings, each a path to input fasta files to be clustered.
reverse_pipe : bool
True/False, whether the reverse reads pipeline is being followed.
Returns
-------
output_fasta_list : list
list of strings, each a path to the output fasta file to which
clusters were written to.
'''
output_fasta_list = []
for input_fasta in input_fasta_list:
output_path = input_fasta.replace('_hits.aln.fa', '_clustered.fa')
cluster_dict = {}
logging.debug('Clustering reads')
if os.path.exists(input_fasta):
reads = self.seqio.read_fasta_file(
input_fasta) # Read in FASTA records
logging.debug('Found %i reads' %
len(reads)) # Report number found
clusters = self.clust.deduplicate(
reads) # Cluster redundant sequences
logging.debug('Clustered to %s groups' %
len(clusters)) # Report number of clusters
logging.debug(
'Writing representative sequences of each cluster to: %s' %
output_path) # Report the name of the file
else:
logging.debug("Found no reads to be clustered")
clusters = []
self.seqio.write_fasta_file(
[x[0] for x in clusters], output_path
) # Choose the first sequence to write to file as representative (all the same anyway)
for cluster in clusters:
cluster_dict[cluster[
0].name] = cluster # assign the cluster to the dictionary
self.seq_library[output_path] = cluster_dict
output_fasta_list.append(output_path)
return output_fasta_list
def __init__(self):
self.clust = Deduplicator()
self.seqio = SequenceIO()
self.seq_library = {}
self.orfm_regex = OrfM.regular_expression()
def setUp(self):
unittest.TestCase.setUp(self)
self.deduplicator = Deduplicator()
self.d = self.deduplicator
def __init__(self):
self.clust = Deduplicator()
self.seqio = SequenceIO()
self.seq_library = {}
self.orfm_regex = OrfM.regular_expression()
class Clusterer:
def __init__(self):
self.clust = Deduplicator()
self.seqio = SequenceIO()
self.seq_library = {}
self.orfm_regex = OrfM.regular_expression()
def uncluster_annotations(self, input_annotations, reverse_pipe):
'''
Update the annotations hash provided by pplacer to include all
representatives within each cluster
Parameters
----------
input_annotations : hash
Classifications for each representative sequence of the clusters.
each key being the sequence name, and the entry being the taxonomy
string as a list.
reverse_pipe : bool
True/False, whether the reverse reads pipeline is being followed.
Returns
-------
output_annotations : hash
An updated version of the above, which includes all reads from
each cluster
'''
output_annotations = {}
for placed_alignment_file_path, clusters in self.seq_library.iteritems():
if reverse_pipe and placed_alignment_file_path.endswith("_reverse_clustered.fa"): continue
placed_alignment_file = os.path.basename(placed_alignment_file_path)
cluster_classifications = input_annotations[placed_alignment_file]
if reverse_pipe:
placed_alignment_base = placed_alignment_file.replace('_forward_clustered.fa', '')
else:
placed_alignment_base = placed_alignment_file.replace('_clustered.fa', '')
output_annotations[placed_alignment_base] = {}
for rep_read_name, rep_read_taxonomy in cluster_classifications.iteritems():
if reverse_pipe:
orfm_regex = OrfM.regular_expression()
clusters={(orfm_regex.match(key).groups(0)[0] if orfm_regex.match(key) else key): item for key, item in clusters.iteritems()}
for read in clusters[rep_read_name]:
output_annotations[placed_alignment_base][read.name] = rep_read_taxonomy
return output_annotations
def cluster(self, input_fasta_list, reverse_pipe):
'''
cluster - Clusters reads at 100% identity level and writes them to
file. Resets the input_fasta variable as the FASTA file containing the
clusters.
Parameters
----------
input_fasta_list : list
list of strings, each a path to input fasta files to be clustered.
reverse_pipe : bool
True/False, whether the reverse reads pipeline is being followed.
Returns
-------
output_fasta_list : list
list of strings, each a path to the output fasta file to which
clusters were written to.
'''
output_fasta_list = []
for input_fasta in input_fasta_list:
output_path = input_fasta.replace('_hits.aln.fa', '_clustered.fa')
cluster_dict = {}
logging.debug('Clustering reads')
if os.path.exists(input_fasta):
reads=self.seqio.read_fasta_file(input_fasta) # Read in FASTA records
logging.debug('Found %i reads' % len(reads)) # Report number found
clusters=self.clust.deduplicate(reads) # Cluster redundant sequences
logging.debug('Clustered to %s groups' % len(clusters)) # Report number of clusters
logging.debug('Writing representative sequences of each cluster to: %s' % output_path) # Report the name of the file
else:
logging.debug("Found no reads to be clustered")
clusters = []
self.seqio.write_fasta_file(
[x[0] for x in clusters],
output_path
) # Choose the first sequence to write to file as representative (all the same anyway)
for cluster in clusters:
cluster_dict[cluster[0].name]=cluster # assign the cluster to the dictionary
self.seq_library[output_path]= cluster_dict
output_fasta_list.append(output_path)
return output_fasta_list
def main(self, **kwargs):
alignment = kwargs.pop('alignment',None)
sequences = kwargs.pop('sequences',None)
taxonomy = kwargs.pop('taxonomy',None)
rerooted_tree = kwargs.pop('rerooted_tree',None)
unrooted_tree = kwargs.pop('unrooted_tree',None)
tree_log = kwargs.pop('tree_log', None)
prefix = kwargs.pop('prefix', None)
rerooted_annotated_tree = kwargs.pop('rerooted_annotated_tree', None)
user_hmm = kwargs.pop('hmm', None)
search_hmm_files = kwargs.pop('search_hmm_files',None)
min_aligned_percent = kwargs.pop('min_aligned_percent',0.01)
taxtastic_taxonomy = kwargs.pop('taxtastic_taxonomy', None)
taxtastic_seqinfo = kwargs.pop('taxtastic_seqinfo', None)
force_overwrite = kwargs.pop('force',False)
graftm_package = kwargs.pop('graftm_package',False)
dereplication_level = kwargs.pop('dereplication_level',False)
threads = kwargs.pop('threads',5)
if len(kwargs) > 0:
raise Exception("Unexpected arguments detected: %s" % kwargs)
seqio = SequenceIO()
locus_name = (os.path.basename(sequences).split('.')[0]
if sequences
else os.path.basename(alignment).split('.')[0])
tmp = tempdir.TempDir()
base = os.path.join(tmp.name, locus_name)
insufficiently_aligned_sequences = [None]
removed_sequence_names = []
tempfiles_to_close = []
if prefix:
output_gpkg_path = prefix
else:
output_gpkg_path = "%s.gpkg" % locus_name
if os.path.exists(output_gpkg_path):
if force_overwrite:
logging.warn("Deleting previous directory %s" % output_gpkg_path)
shutil.rmtree(output_gpkg_path)
else:
raise Exception("Cowardly refusing to overwrite gpkg to already existing %s" % output_gpkg_path)
logging.info("Building gpkg for %s" % output_gpkg_path)
# Read in taxonomy somehow
gtns = Getaxnseq()
if rerooted_annotated_tree:
logging.info("Building seqinfo and taxonomy file from input annotated tree")
taxonomy_definition = TaxonomyExtractor().taxonomy_from_annotated_tree(\
Tree.get(path=rerooted_annotated_tree, schema='newick'))
elif taxonomy:
logging.info("Building seqinfo and taxonomy file from input taxonomy")
taxonomy_definition = GreenGenesTaxonomy.read_file(taxonomy).taxonomy
elif taxtastic_seqinfo and taxtastic_taxonomy:
logging.info("Reading taxonomy from taxtastic taxonomy and seqinfo files")
taxonomy_definition = gtns.read_taxtastic_taxonomy_and_seqinfo\
(open(taxtastic_taxonomy),
open(taxtastic_seqinfo))
else:
raise Exception("Taxonomy is required somehow e.g. by --taxonomy or --rerooted_annotated_tree")
# Check for duplicates
logging.info("Checking for duplicate sequences")
dup = self._check_for_duplicate_sequence_names(sequences)
if dup:
raise Exception("Found duplicate sequence name '%s' in sequences input file" % dup)
output_alignment_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.aln.faa')
tempfiles_to_close.append(output_alignment_fh)
output_alignment = output_alignment_fh.name
if user_hmm:
align_hmm = user_hmm
else:
align_hmm_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='_align.hmm')
tempfiles_to_close.append(align_hmm_fh)
align_hmm = align_hmm_fh.name
if alignment:
dup = self._check_for_duplicate_sequence_names(alignment)
if dup:
raise Exception("Found duplicate sequence name '%s' in alignment input file" % dup)
ptype = self._get_hmm_from_alignment(alignment,
align_hmm,
output_alignment)
else:
logging.info("Aligning sequences to create aligned FASTA file")
ptype, output_alignment = self._align_and_create_hmm(sequences, alignment, user_hmm,
align_hmm, output_alignment, threads)
logging.info("Checking for incorrect or fragmented reads")
insufficiently_aligned_sequences = self._check_reads_hit(open(output_alignment),
min_aligned_percent)
while len(insufficiently_aligned_sequences) > 0:
logging.warn("One or more alignments do not span > %.2f %% of HMM" % (min_aligned_percent*100))
for s in insufficiently_aligned_sequences:
logging.warn("Insufficient alignment of %s, not including this sequence" % s)
sequences2_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.faa')
tempfiles_to_close.append(sequences2_fh)
sequences2 = sequences2_fh.name
num_sequences = self._remove_sequences_from_alignment(insufficiently_aligned_sequences,
sequences,
sequences2)
sequences = sequences2
if alignment:
alignment2_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.aln.faa')
tempfiles_to_close.append(alignment2_fh)
alignment2 = alignment2_fh.name
num_sequences = self._remove_sequences_from_alignment(insufficiently_aligned_sequences,
alignment,
alignment2)
alignment = alignment2
for name in insufficiently_aligned_sequences:
if rerooted_tree or rerooted_annotated_tree:
logging.warning('''Sequence %s in provided alignment does not meet the --min_aligned_percent cutoff. This sequence will be removed from the tree
in the final GraftM package. If you are sure these sequences are correct, turn off the --min_aligned_percent cutoff, provide it with a 0 (e.g. --min_aligned_percent 0) ''' % name)
removed_sequence_names.append(name)
logging.info("After removing %i insufficiently aligned sequences, left with %i sequences" % (len(insufficiently_aligned_sequences), num_sequences))
if num_sequences < 4:
raise Exception("Too few sequences remaining in alignment after removing insufficiently aligned sequences: %i" % num_sequences)
else:
logging.info("Reconstructing the alignment and HMM from remaining sequences")
output_alignment_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.aln.faa')
tempfiles_to_close.append(output_alignment_fh)
output_alignment = output_alignment_fh.name
if not user_hmm:
align_hmm_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.hmm')
tempfiles_to_close.append(align_hmm_fh)
align_hmm = align_hmm_fh.name
ptype, output_alignment= self._align_and_create_hmm(sequences, alignment, user_hmm,
align_hmm, output_alignment, threads)
logging.info("Checking for incorrect or fragmented reads")
insufficiently_aligned_sequences = self._check_reads_hit(open(output_alignment),
min_aligned_percent)
if not search_hmm_files:
search_hmm_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='_search.hmm')
tempfiles_to_close.append(search_hmm_fh)
search_hmm = search_hmm_fh.name
self._create_search_hmm(sequences, taxonomy_definition, search_hmm, dereplication_level, threads)
search_hmm_files = [search_hmm]
# Make sure each sequence has been assigned a taxonomy:
aligned_sequence_objects = seqio.read_fasta_file(output_alignment)
unannotated = []
for s in aligned_sequence_objects:
if s.name not in taxonomy_definition:
unannotated.append(s.name)
if len(unannotated) > 0:
for s in unannotated:
logging.error("Unable to find sequence '%s' in the taxonomy definition" % s)
raise Exception("All sequences must be assigned a taxonomy, cannot continue")
logging.debug("Looking for non-standard characters in aligned sequences")
self._mask_strange_sequence_letters(aligned_sequence_objects, ptype)
# Deduplicate sequences - pplacer cannot handle these
logging.info("Deduplicating sequences")
dedup = Deduplicator()
deduplicated_arrays = dedup.deduplicate(aligned_sequence_objects)
deduplicated_taxonomy = dedup.lca_taxonomy(deduplicated_arrays, taxonomy_definition)
deduplicated_taxonomy_hash = {}
for i, tax in enumerate(deduplicated_taxonomy):
deduplicated_taxonomy_hash[deduplicated_arrays[i][0].name] = tax
deduplicated_alignment_file = base+"_deduplicated_aligned.fasta"
seqio.write_fasta_file([seqs[0] for seqs in deduplicated_arrays],
deduplicated_alignment_file)
logging.info("Removed %i sequences as duplicates, leaving %i non-identical sequences"\
% ((len(aligned_sequence_objects)-len(deduplicated_arrays)),
len(deduplicated_arrays)))
# Get corresponding unaligned sequences
filtered_names=[]
for list in [x for x in [x[1:] for x in deduplicated_arrays] if x]:
for seq in list:
filtered_names.append(seq.name)
sequences2_fh = tempfile.NamedTemporaryFile(prefix='graftm', suffix='.faa')
tempfiles_to_close.append(sequences2_fh)
sequences2 = sequences2_fh.name
# Create tree unless one was provided
if not rerooted_tree and not rerooted_annotated_tree and not unrooted_tree:
logging.debug("No tree provided")
logging.info("Building tree")
log_file, tre_file = self._build_tree(deduplicated_alignment_file,
base, ptype,
self.fasttree)
no_reroot = False
else:
if rerooted_tree:
logging.debug("Found unannotated pre-rerooted tree file %s" % rerooted_tree)
tre_file=rerooted_tree
no_reroot = True
elif rerooted_annotated_tree:
logging.debug("Found annotated pre-rerooted tree file %s" % rerooted_tree)
tre_file=rerooted_annotated_tree
no_reroot = True
elif unrooted_tree:
logging.info("Using input unrooted tree")
tre_file = unrooted_tree
no_reroot = False
else:
raise
# Remove any sequences from the tree that are duplicates
cleaner = DendropyTreeCleaner()
tree = Tree.get(path=tre_file, schema='newick')
for group in deduplicated_arrays:
[removed_sequence_names.append(s.name) for s in group[1:]]
cleaner.remove_sequences(tree, removed_sequence_names)
# Ensure there is nothing amiss now as a user-interface thing
cleaner.match_alignment_and_tree_sequence_ids(\
[g[0].name for g in deduplicated_arrays], tree)
if tree_log:
# User specified a log file, go with that
logging.debug("Using user-specified log file %s" % tree_log)
log_file = tree_log
else:
logging.info("Generating log file")
log_file_tempfile = tempfile.NamedTemporaryFile(suffix='.tree_log', prefix='graftm')
tempfiles_to_close.append(log_file_tempfile)
log_file = log_file_tempfile.name
tre_file_tempfile = tempfile.NamedTemporaryFile(suffix='.tree', prefix='graftm')
tempfiles_to_close.append(tre_file_tempfile)
tre_file = tre_file_tempfile.name
with tempfile.NamedTemporaryFile(suffix='.tree', prefix='graftm') as f:
# Make the newick file simple (ie. un-arb it) for fasttree.
cleaner.write_fasttree_newick(tree, f)
f.flush()
self._generate_tree_log_file(f.name, deduplicated_alignment_file,
tre_file, log_file, ptype, self.fasttree)
# Create tax and seqinfo .csv files
taxonomy_to_keep=[
seq.name for seq in
[x for x in [x[0] for x in deduplicated_arrays]
if x]
]
refpkg = "%s.refpkg" % output_gpkg_path
self.the_trash.append(refpkg)
if taxtastic_taxonomy and taxtastic_seqinfo:
logging.info("Creating reference package")
refpkg = self._taxit_create(base, deduplicated_alignment_file,
tre_file, log_file, taxtastic_taxonomy,
taxtastic_seqinfo, refpkg, no_reroot)
else:
gtns = Getaxnseq()
seq = base+"_seqinfo.csv"
tax = base+"_taxonomy.csv"
self.the_trash += [seq, tax]
if rerooted_annotated_tree:
logging.info("Building seqinfo and taxonomy file from input annotated tree")
taxonomy_definition = TaxonomyExtractor().taxonomy_from_annotated_tree(
Tree.get(path=rerooted_annotated_tree, schema='newick'))
elif taxonomy:
logging.info("Building seqinfo and taxonomy file from input taxonomy")
taxonomy_definition = GreenGenesTaxonomy.read_file(taxonomy).taxonomy
else:
raise Exception("Programming error: Taxonomy is required somehow e.g. by --taxonomy or --rerooted_annotated_tree")
taxonomy_definition = {x:taxonomy_definition[x]
for x in taxonomy_definition
if x in taxonomy_to_keep}
gtns.write_taxonomy_and_seqinfo_files(taxonomy_definition,
tax,
seq)
# Create the reference package
logging.info("Creating reference package")
refpkg = self._taxit_create(base, deduplicated_alignment_file,
tre_file, log_file, tax, seq, refpkg,
no_reroot)
if sequences:
# Run diamond makedb
logging.info("Creating diamond database")
if ptype == Create._PROTEIN_PACKAGE_TYPE:
cmd = "diamond makedb --in '%s' -d '%s'" % (sequences, base)
extern.run(cmd)
diamondb = '%s.dmnd' % base
elif ptype == Create._NUCLEOTIDE_PACKAGE_TYPE:
diamondb = None
else: raise Exception("Programming error")
else:
diamondb = None
if sequences:
# Get range
max_range = self._define_range(sequences)
else:
max_range = self._define_range(alignment)
# Compile the gpkg
logging.info("Compiling gpkg")
GraftMPackageVersion3.compile(output_gpkg_path, refpkg, align_hmm, diamondb,
max_range, sequences, search_hmm_files=search_hmm_files)
logging.info("Cleaning up")
self._cleanup(self.the_trash)
for tf in tempfiles_to_close:
tf.close()
# Test out the gpkg just to be sure.
#
# TODO: Use graftM through internal means rather than via extern. This
# requires some refactoring so that graft() can be called easily with
# sane defaults.
logging.info("Testing gpkg package works")
self._test_package(output_gpkg_path)
logging.info("Finished\n")
def main(self, **kwargs):
alignment = kwargs.pop('alignment', None)
sequences = kwargs.pop('sequences', None)
taxonomy = kwargs.pop('taxonomy', None)
rerooted_tree = kwargs.pop('rerooted_tree', None)
unrooted_tree = kwargs.pop('unrooted_tree', None)
tree_log = kwargs.pop('tree_log', None)
prefix = kwargs.pop('prefix', None)
rerooted_annotated_tree = kwargs.pop('rerooted_annotated_tree', None)
user_hmm = kwargs.pop('hmm', None)
search_hmm_files = kwargs.pop('search_hmm_files', None)
min_aligned_percent = kwargs.pop('min_aligned_percent', 0.01)
taxtastic_taxonomy = kwargs.pop('taxtastic_taxonomy', None)
taxtastic_seqinfo = kwargs.pop('taxtastic_seqinfo', None)
force_overwrite = kwargs.pop('force', False)
graftm_package = kwargs.pop('graftm_package', False)
dereplication_level = kwargs.pop('dereplication_level', False)
threads = kwargs.pop('threads', 5)
if len(kwargs) > 0:
raise Exception("Unexpected arguments detected: %s" % kwargs)
seqio = SequenceIO()
locus_name = (os.path.basename(sequences).split('.')[0] if sequences
else os.path.basename(alignment).split('.')[0])
tmp = tempdir.TempDir()
base = os.path.join(tmp.name, locus_name)
insufficiently_aligned_sequences = [None]
removed_sequence_names = []
if prefix:
output_gpkg_path = prefix
else:
output_gpkg_path = "%s.gpkg" % locus_name
if os.path.exists(output_gpkg_path):
if force_overwrite:
logging.warn("Deleting previous directory %s" %
output_gpkg_path)
shutil.rmtree(output_gpkg_path)
else:
raise Exception(
"Cowardly refusing to overwrite gpkg to already existing %s"
% output_gpkg_path)
logging.info("Building gpkg for %s" % output_gpkg_path)
# Read in taxonomy somehow
gtns = Getaxnseq()
if rerooted_annotated_tree:
logging.info(
"Building seqinfo and taxonomy file from input annotated tree")
taxonomy_definition = TaxonomyExtractor().taxonomy_from_annotated_tree(\
Tree.get(path=rerooted_annotated_tree, schema='newick'))
elif taxonomy:
logging.info(
"Building seqinfo and taxonomy file from input taxonomy")
taxonomy_definition = GreenGenesTaxonomy.read_file(
taxonomy).taxonomy
elif taxtastic_seqinfo and taxtastic_taxonomy:
logging.info(
"Reading taxonomy from taxtastic taxonomy and seqinfo files")
taxonomy_definition = gtns.read_taxtastic_taxonomy_and_seqinfo\
(open(taxtastic_taxonomy),
open(taxtastic_seqinfo))
else:
raise Exception(
"Taxonomy is required somehow e.g. by --taxonomy or --rerooted_annotated_tree"
)
# Check for duplicates
logging.info("Checking for duplicate sequences")
dup = self._check_for_duplicate_sequence_names(sequences)
if dup:
raise Exception(
"Found duplicate sequence name '%s' in sequences input file" %
dup)
output_alignment = tempfile.NamedTemporaryFile(prefix='graftm',
suffix='.aln.faa').name
align_hmm = (user_hmm if user_hmm else tempfile.NamedTemporaryFile(
prefix='graftm', suffix='_align.hmm').name)
if alignment:
dup = self._check_for_duplicate_sequence_names(alignment)
if dup:
raise Exception(
"Found duplicate sequence name '%s' in alignment input file"
% dup)
ptype = self._get_hmm_from_alignment(alignment, align_hmm,
output_alignment)
else:
logging.info("Aligning sequences to create aligned FASTA file")
ptype, output_alignment = self._align_and_create_hmm(
sequences, alignment, user_hmm, align_hmm, output_alignment,
threads)
logging.info("Checking for incorrect or fragmented reads")
insufficiently_aligned_sequences = self._check_reads_hit(
open(output_alignment), min_aligned_percent)
while len(insufficiently_aligned_sequences) > 0:
logging.warn(
"One or more alignments do not span > %.2f %% of HMM" %
(min_aligned_percent * 100))
for s in insufficiently_aligned_sequences:
logging.warn(
"Insufficient alignment of %s, not including this sequence"
% s)
_, sequences2 = tempfile.mkstemp(prefix='graftm', suffix='.faa')
num_sequences = self._remove_sequences_from_alignment(
insufficiently_aligned_sequences, sequences, sequences2)
sequences = sequences2
if alignment:
_, alignment2 = tempfile.mkstemp(prefix='graftm',
suffix='.aln.faa')
num_sequences = self._remove_sequences_from_alignment(
insufficiently_aligned_sequences, alignment, alignment2)
alignment = alignment2
for name in insufficiently_aligned_sequences:
if rerooted_tree or rerooted_annotated_tree:
logging.warning(
'''Sequence %s in provided alignment does not meet the --min_aligned_percent cutoff. This sequence will be removed from the tree
in the final GraftM package. If you are sure these sequences are correct, turn off the --min_aligned_percent cutoff, provide it with a 0 (e.g. --min_aligned_percent 0) '''
% name)
removed_sequence_names.append(name)
logging.info(
"After removing %i insufficiently aligned sequences, left with %i sequences"
% (len(insufficiently_aligned_sequences), num_sequences))
if num_sequences < 4:
raise Exception(
"Too few sequences remaining in alignment after removing insufficiently aligned sequences: %i"
% num_sequences)
else:
logging.info(
"Reconstructing the alignment and HMM from remaining sequences"
)
output_alignment = tempfile.NamedTemporaryFile(
prefix='graftm', suffix='.aln.faa').name
if not user_hmm:
align_hmm = tempfile.NamedTemporaryFile(prefix='graftm',
suffix='.hmm').name
ptype, output_alignment = self._align_and_create_hmm(
sequences, alignment, user_hmm, align_hmm,
output_alignment, threads)
logging.info("Checking for incorrect or fragmented reads")
insufficiently_aligned_sequences = self._check_reads_hit(
open(output_alignment), min_aligned_percent)
if not search_hmm_files:
search_hmm = tempfile.NamedTemporaryFile(prefix='graftm',
suffix='_search.hmm').name
self._create_search_hmm(sequences, taxonomy_definition, search_hmm,
dereplication_level, threads)
search_hmm_files = [search_hmm]
# Make sure each sequence has been assigned a taxonomy:
aligned_sequence_objects = seqio.read_fasta_file(output_alignment)
unannotated = []
for s in aligned_sequence_objects:
if s.name not in taxonomy_definition:
unannotated.append(s.name)
if len(unannotated) > 0:
for s in unannotated:
logging.error(
"Unable to find sequence '%s' in the taxonomy definition" %
s)
raise Exception(
"All sequences must be assigned a taxonomy, cannot continue")
logging.debug(
"Looking for non-standard characters in aligned sequences")
self._mask_strange_sequence_letters(aligned_sequence_objects, ptype)
# Deduplicate sequences - pplacer cannot handle these
logging.info("Deduplicating sequences")
dedup = Deduplicator()
deduplicated_arrays = dedup.deduplicate(aligned_sequence_objects)
deduplicated_taxonomy = dedup.lca_taxonomy(deduplicated_arrays,
taxonomy_definition)
deduplicated_taxonomy_hash = {}
for i, tax in enumerate(deduplicated_taxonomy):
deduplicated_taxonomy_hash[deduplicated_arrays[i][0].name] = tax
deduplicated_alignment_file = base + "_deduplicated_aligned.fasta"
seqio.write_fasta_file([seqs[0] for seqs in deduplicated_arrays],
deduplicated_alignment_file)
logging.info("Removed %i sequences as duplicates, leaving %i non-identical sequences"\
% ((len(aligned_sequence_objects)-len(deduplicated_arrays)),
len(deduplicated_arrays)))
# Get corresponding unaligned sequences
filtered_names = []
for list in [x for x in [x[1:] for x in deduplicated_arrays] if x]:
for seq in list:
filtered_names.append(seq.name)
_, sequences2 = tempfile.mkstemp(prefix='graftm', suffix='.faa')
# Create tree unless one was provided
if not rerooted_tree and not rerooted_annotated_tree and not unrooted_tree:
logging.debug("No tree provided")
logging.info("Building tree")
log_file, tre_file = self._build_tree(deduplicated_alignment_file,
base, ptype, self.fasttree)
no_reroot = False
else:
if rerooted_tree:
logging.debug("Found unannotated pre-rerooted tree file %s" %
rerooted_tree)
tre_file = rerooted_tree
no_reroot = True
elif rerooted_annotated_tree:
logging.debug("Found annotated pre-rerooted tree file %s" %
rerooted_tree)
tre_file = rerooted_annotated_tree
no_reroot = True
elif unrooted_tree:
logging.info("Using input unrooted tree")
tre_file = unrooted_tree
no_reroot = False
else:
raise
# Remove any sequences from the tree that are duplicates
cleaner = DendropyTreeCleaner()
tree = Tree.get(path=tre_file, schema='newick')
for group in deduplicated_arrays:
[removed_sequence_names.append(s.name) for s in group[1:]]
cleaner.remove_sequences(tree, removed_sequence_names)
# Ensure there is nothing amiss now as a user-interface thing
cleaner.match_alignment_and_tree_sequence_ids(\
[g[0].name for g in deduplicated_arrays], tree)
if tree_log:
# User specified a log file, go with that
logging.debug("Using user-specified log file %s" % tree_log)
log_file = tree_log
else:
logging.info("Generating log file")
log_file_tempfile = tempfile.NamedTemporaryFile(
suffix='.tree_log', prefix='graftm')
log_file = log_file_tempfile.name
tre_file_tempfile = tempfile.NamedTemporaryFile(
suffix='.tree', prefix='graftm')
tre_file = tre_file_tempfile.name
with tempfile.NamedTemporaryFile(suffix='.tree',
prefix='graftm') as f:
# Make the newick file simple (ie. un-arb it) for fasttree.
cleaner.write_fasttree_newick(tree, f)
f.flush()
self._generate_tree_log_file(f.name,
deduplicated_alignment_file,
tre_file, log_file, ptype,
self.fasttree)
# Create tax and seqinfo .csv files
taxonomy_to_keep = [
seq.name
for seq in [x for x in [x[0] for x in deduplicated_arrays] if x]
]
refpkg = "%s.refpkg" % output_gpkg_path
self.the_trash.append(refpkg)
if taxtastic_taxonomy and taxtastic_seqinfo:
logging.info("Creating reference package")
refpkg = self._taxit_create(base, deduplicated_alignment_file,
tre_file, log_file, taxtastic_taxonomy,
taxtastic_seqinfo, refpkg, no_reroot)
else:
gtns = Getaxnseq()
seq = base + "_seqinfo.csv"
tax = base + "_taxonomy.csv"
self.the_trash += [seq, tax]
if rerooted_annotated_tree:
logging.info(
"Building seqinfo and taxonomy file from input annotated tree"
)
taxonomy_definition = TaxonomyExtractor(
).taxonomy_from_annotated_tree(
Tree.get(path=rerooted_annotated_tree, schema='newick'))
elif taxonomy:
logging.info(
"Building seqinfo and taxonomy file from input taxonomy")
taxonomy_definition = GreenGenesTaxonomy.read_file(
taxonomy).taxonomy
else:
raise Exception(
"Programming error: Taxonomy is required somehow e.g. by --taxonomy or --rerooted_annotated_tree"
)
taxonomy_definition = {
x: taxonomy_definition[x]
for x in taxonomy_definition if x in taxonomy_to_keep
}
gtns.write_taxonomy_and_seqinfo_files(taxonomy_definition, tax,
seq)
# Create the reference package
logging.info("Creating reference package")
refpkg = self._taxit_create(base, deduplicated_alignment_file,
tre_file, log_file, tax, seq, refpkg,
no_reroot)
if sequences:
# Run diamond makedb
logging.info("Creating diamond database")
if ptype == Create._PROTEIN_PACKAGE_TYPE:
cmd = "diamond makedb --in '%s' -d '%s'" % (sequences, base)
extern.run(cmd)
diamondb = '%s.dmnd' % base
elif ptype == Create._NUCLEOTIDE_PACKAGE_TYPE:
diamondb = None
else:
raise Exception("Programming error")
else:
diamondb = None
if sequences:
# Get range
max_range = self._define_range(sequences)
else:
max_range = self._define_range(alignment)
# Compile the gpkg
logging.info("Compiling gpkg")
GraftMPackageVersion3.compile(output_gpkg_path,
refpkg,
align_hmm,
diamondb,
max_range,
sequences,
search_hmm_files=search_hmm_files)
logging.info("Cleaning up")
self._cleanup(self.the_trash)
# Test out the gpkg just to be sure.
#
# TODO: Use graftM through internal means rather than via extern. This
# requires some refactoring so that graft() can be called easily with
# sane defaults.
logging.info("Testing gpkg package works")
self._test_package(output_gpkg_path)
logging.info("Finished\n")