diff options
Diffstat (limited to 'workflows/pangenome-generate/dups2metadata.py')
| -rw-r--r-- | workflows/pangenome-generate/dups2metadata.py | 17 |
1 files changed, 17 insertions, 0 deletions
diff --git a/workflows/pangenome-generate/dups2metadata.py b/workflows/pangenome-generate/dups2metadata.py new file mode 100644 index 0000000..9bda10a --- /dev/null +++ b/workflows/pangenome-generate/dups2metadata.py @@ -0,0 +1,17 @@ +import sys + +md = open(sys.argv[1], "rt") +for d in md: + print(d) + +if len(sys.argv) < 3: + exit(0) + +sameseqs = open(sys.argv[2], "rt") +for d in sameseqs: + logging.warn(d) + g = re.match(r"\d+\t(.*)", d) + logging.warn("%s", g.group(1)) + sp = g.group(1).split(",") + for n in sp[1:]: + print("<%s> <http://biohackathon.org/bh20-seq-schema/has_duplicate_sequence> <%s> ." % (n.strip(), sp[0].strip())) |