def offtarget(organism,
offtarget_databases,
offtarget_names,
tmp_dir=None):
if not tmp_dir:
tmp_dir = "/data/organismos/" + organism + "/annotation/"
mkdir(tmp_dir)
proteins = tmp_dir + "proteins.fasta"
if not os.path.exists(proteins):
BioMongoDB.protein_fasta(proteins, organism)
results = Offtarget.offtargets(proteins, tmp_dir, offtarget_databases)
for i, name in enumerate(offtarget_names):
load_blast_features(organism,
results[i],
name,
min_identity=0.4,
min_query_coverage=0.4,
min_hit_coverage=0.4)
("MpyloriIndia", "Helicobacter pylori India7 (e-proteobacteria)",
"/data/organismos/MpyloriIndia/GCF_000185185.1_ASM18518v1_genomic.gbff",
907238),
]
for name, org, ann_path, tax in orgs:
organism = name
mkdir("/data/organismos/" + name + "/annotation/offtarget")
mkdir("/data/organismos/" + name + "/annotation/pwtools")
mkdir("/data/organismos/" + name + "/annotation/pathways")
mkdir("/data/organismos/" + name + "/estructura/raw")
mkdir("/data/organismos/" + name + "/estructura/sndg/modelos")
mkdir("/data/organismos/" + name + "/estructura/sndg/pockets")
from_ref_seq(name, ann_path, tax=tax, cpus=3)
mdb.protein_fasta("/data/organismos/" + name + "/annotation/proteins.faa",
name)
update_proteins("/tmp/" + name + "/",
"/data/organismos/" + name + "/annotation/proteins.faa",
name,
1003200,
db_init=mysqldb)
Offtargeting.offtargets(
"/data/organismos/" + name + "/annotation/proteins.faa",
"/data/organismos/" + name + "/annotation/offtarget/",
offtarget_dbs=[
"/data/databases/deg/degaa-p.dat",
"/data/databases/human/gencode.v17.pc_translations.fa",
"/data/databases/human/gut_microbiota.fasta"
])
import_prop_blast(mdb.db,
"options": ["No", "Yes"],
"description": "Has a hit in Database of Essential Genes"
})
}
from SNDG.Sequence import read_blast_table
from tqdm import tqdm
# cols = list(SeqCollection.objects(name__nin=["cruzi","pdb"]))
cols = list(SeqCollection.objects(name__nin=["cruzi", "pdb"]))
cpus = 4
db = mdb.db
for seqCol in tqdm(cols):
mkdir("/data/organismos/" + seqCol.name + "/contigs")
proteome = "/data/organismos/" + seqCol.name + "/contigs/genoma.fasta"
if not os.path.exists(proteome):
mdb.protein_fasta(proteome, seqCol.name)
out = "/data/organismos/" + seqCol.name + "/annotation/offtarget/"
mkdir(out)
if not seqCol.has_druggability_param("human_offtarget"):
seqCol.druggabilityParams.append(off_props["human_offtarget"])
db = "/data/databases/human/gencode.v17.pc_translations.fa"
execute(
"blastp -evalue 1e-5 -max_hsps 1 -outfmt 6 -max_target_seqs 1 -db {db} -query {query} -out {out} -num_threads {cpus}",
db=db,
query=proteome,
out=out + "human_offtarget.tbl",
cpus=cpus)