Call clustalw to compare sequence to reference

Arvados-DCO-1.1-Signed-off-by: Peter Amstutz <peter.amstutz@curii.com>

1 files changed, 55 insertions, 8 deletions

diff --git a/bh20sequploader/qc_fasta.py b/bh20sequploader/qc_fasta.py
index e47d66b..2600cfa 100644
--- a/bh20sequploader/qc_fasta.py
+++ b/bh20sequploader/qc_fasta.py
@@ -1,6 +1,25 @@
 import pkg_resources
 import tempfile
 import magic
+import subprocess
+import tempfile
+import logging
+import re
+
+def read_fasta(sequence):
+    entries = 0
+    bases = []
+    label = None
+    for line in sequence:
+        if line.startswith(">"):
+            label = line
+            entries += 1
+        else:
+            bases.append(line)
+        if entries > 1:
+            raise ValueError("FASTA file contains multiple entries")
+            break
+    return label, bases
 
 def qc_fasta(sequence):
     schema_resource = pkg_resources.resource_stream(__name__, "validation/formats")
@@ -13,16 +32,44 @@ def qc_fasta(sequence):
     sequence.seek(0)
     if seq_type == "text/fasta":
         # ensure that contains only one entry
-        entries = 0
-        for line in sequence:
-            if line.startswith(">"):
-                entries += 1
-            if entries > 1:
-                raise ValueError("FASTA file contains multiple entries")
-                break
+        submitlabel, submitseq = read_fasta(sequence)
         sequence.seek(0)
+
+        with tempfile.NamedTemporaryFile() as tmp1:
+            refstring = pkg_resources.resource_string(__name__, "SARS-CoV-2-reference.fasta")
+            tmp1.write(refstring)
+            tmp1.write(submitlabel.encode("utf8"))
+            tmp1.write(("".join(submitseq)).encode("utf8"))
+            tmp1.flush()
+            try:
+                cmd = ["clustalw", "-infile="+tmp1.name,
+                       "-quicktree", "-iteration=none", "-type=DNA"]
+                print("QC checking similarity to reference")
+                print(" ".join(cmd))
+                result = subprocess.run(cmd, capture_output=True)
+                res = result.stdout.decode("utf-8")
+                g1 = re.search(r"^Sequence 1: [^ ]+ +(\d+) bp$", res, flags=re.MULTILINE)
+                refbp = float(g1.group(1))
+                g2 = re.search(r"^Sequence 2: [^ ]+ +(\d+) bp$", res, flags=re.MULTILINE)
+                subbp = float(g2.group(1))
+                g3 = re.search(r"^Sequences \(1:2\) Aligned\. Score: (\d+(\.\d+)?)$", res, flags=re.MULTILINE)
+                similarity = float(g3.group(1))
+
+                print(g1.group(0))
+                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)
+
+            if (subbp/refbp) < .7:
+                raise ValueError("QC fail: submit sequence length is shorter than 70% reference")
+            if (subbp/refbp) > 1.3:
+                raise ValueError("QC fail: submit sequence length is greater than 130% reference")
+            if similarity < 70.0:
+                raise ValueError("QC fail: submit similarity is less than 70%")
+
         return "sequence.fasta"
     elif seq_type == "text/fastq":
         return "reads.fastq"
     else:
-        raise ValueError("Sequence file does not look like FASTA or FASTQ")
+        raise ValueError("Sequence file does not look like a DNA FASTA or FASTQ")