seqordered = [seq for seq in sorted(seqs, key=lambda x: x.id, reverse=True)]
pairs = []
unpaired = []
first = None
nxt = None
while seqordered:
if (first is None):
first = seqordered.pop()
if (seqordered):
if (nxt is None):
nxt = seqordered.pop()
if first.id == nxt.id:
pairs.append(pair.Pair([first, nxt]))
first = None
nxt = None
else:
unpaired.append(pair.Pair([first]))
first = nxt
nxt = None
else:
unpaired.append(pair.Pair([first]))
first = None
print("Processed", len(pairs), "pairs of sequences")
print("and ", len(unpaired), " unpaired sequences")
# Extracted from Bryan's code to fix output_schief_csv()
cells[k] = {'heavy': [], 'light': []}
for s in seqordered:
if (s['chain'] == 'heavy'):
cells[s.id]['heavy'].append(s)
else:
cells[s.id]['light'].append(s)
# Pair sorted sequences based on munged id.
# Unpaired sequences aren't exported. Should they be?
pairs = []
unpaired = []
# Within a 'cluster' of sequences we want to pair each heavy and each light chain sequence.
for cell in cells:
if (len(cells[cell]['heavy']) == 0) and (len(cells[cell]['light']) > 0):
unpaired.extend([pair.Pair([s]) for s in cells[cell]['light']])
if (len(cells[cell]['light']) == 0) and (len(cells[cell]['heavy']) > 0):
unpaired.extend([pair.Pair([s]) for s in cells[cell]['heavy']])
if (len(cells[cell]['light']) > 0) and (len(cells[cell]['heavy']) > 0):
# for each heavy pair it with each light.
for heavy in cells[cell]['heavy']:
for light in cells[cell]['light']:
pairs.append(pair.Pair([heavy, light]))
print("Processed", len(pairs), "pairs of sequences")
print("and ", len(unpaired), " unpaired sequences")
# Extracted from Bryan's code to fix output_schief_csv()
# FROM https://github.com/briney/abtools/blob/master/abtools/pipeline.py
from abutils.core import sequence
from abutils.core import pair
import pandas as pd
import numpy as np
import json
from pandas.io.json import json_normalize
import sys, os
pairs = []
with open(sys.argv[1]) as f:
for line in f:
d = json.loads(line.strip())
# Bryan has a pair object
seq = sequence.Sequence(d)
seqs = [seq]
abpair = pair.Pair(seqs)
pairs.append(abpair)
# Built in json flatten from pandas does an OK job.
#json_df = json_normalize(d)
#json_df.to_csv("dump.csv", sep=',')
# An example of a mole elaborate json flatten that we can repurpose
# From https://towardsdatascience.com/flattening-json-objects-in-python-f5343c794b10
def flatten_json(y):
out = {}
def flatten(x, name=''):
if type(x) is dict:
for a in x: