diff options
| author | Peter Amstutz | 2020-07-03 17:08:25 -0400 |
|---|---|---|
| committer | Peter Amstutz | 2020-07-03 17:08:25 -0400 |
| commit | 04df498f5cd85015afce79e1e87a3979e596dcc6 (patch) | |
| tree | 98f7e65d10b41dc438189f704aedde13133a77a7 /bh20sequploader/qc_fasta.py | |
| parent | c50e611f7be281a3b7955254be097c01e6461c20 (diff) | |
| download | bh20-seq-resource-04df498f5cd85015afce79e1e87a3979e596dcc6.tar.gz bh20-seq-resource-04df498f5cd85015afce79e1e87a3979e596dcc6.tar.lz bh20-seq-resource-04df498f5cd85015afce79e1e87a3979e596dcc6.zip | |
Adding --skip-qc
Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>
Diffstat (limited to 'bh20sequploader/qc_fasta.py')
| -rw-r--r-- | bh20sequploader/qc_fasta.py | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py index 1587def..944b52c 100644 --- a/bh20sequploader/qc_fasta.py +++ b/bh20sequploader/qc_fasta.py @@ -25,7 +25,7 @@ def read_fasta(sequence): raise ValueError("FASTA file contains multiple entries") return label, bases -def qc_fasta(arg_sequence): +def qc_fasta(arg_sequence, check_with_clustalw=True): log.debug("Starting qc_fasta") schema_resource = pkg_resources.resource_stream(__name__, "validation/formats") with tempfile.NamedTemporaryFile() as tmp: @@ -64,6 +64,9 @@ def qc_fasta(arg_sequence): refbp = 0 similarity = 0 try: + if not check_with_clustalw: + raise Exception("skipping QC") + cmd = ["clustalw", "-infile="+tmp1.name, "-quicktree", "-iteration=none", "-type=DNA"] print("QC checking similarity to reference") @@ -81,7 +84,7 @@ def qc_fasta(arg_sequence): print(g2.group(0)) print(g3.group(0)) except Exception as e: - logging.warn("Error trying to QC against reference sequence using 'clustalw': %s", e) + logging.warn("QC against reference sequence using 'clustalw': %s", e) if refbp and (subbp/refbp) < .7: raise ValueError("QC fail: submit sequence length is shorter than 70% reference") |