scNLP/RunSCNLP.py at main · alexcwsmith/scNLP · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sun Jul  5 17:08:37 2020

@author: smith
"""

import os
os.chdir("/home/smith/scNLP/")
from config.PubMedScraperUtils import *
import singleCellNLP as scn
from scipy import stats
import statsmodels.api as sm

# If you want to update the manifest file first:
clu = int(cluster.strip('Cluster'))
clusters = ["Cluster" + str(x) for x in range(clu)]
maniDf = scn.updateManifest(clusters, rettype="full")

# Run NLP on current cluster:
scn.runNLP(cluster, rettype="full", n_genes=25, copy=False, dlProcesses=3)

# Downstream analysis:
category = "Physio"  # NER recognized categories: Physio, Label (region), NT, CellType
#types = scn.concatFrequencies(clusters, category=category, save=True)
types = scn.concatFrequencies(clusters, category=category, save=True)
catTypes = scn.filterConcatFrequencies(
    types, category=category, min_count=50, save=True
)

cat = pd.DataFrame()
for cluster in clusters:
    min_count = 300
    lz = scn.findUpregulatedEntities(cluster, category=category, min_count=min_count)
    resDf = pd.DataFrame(lz)
    resDf.columns = [
        cluster + " term",
        cluster + "_pvalue",
        cluster + "_corrected_p",
        cluster + "_oddsratio",
    ]
    resDf = resDf.loc[resDf[cluster + "_oddsratio"] > 1.0]
    resDf = resDf.sort_values(by=cluster + "_corrected_p", ascending=True)
    cat = pd.concat([cat, resDf], axis=1)
cat.to_excel(
    os.path.join(
        resultDirectory,
        "Combined_Clusters_Enriched_"
        + category
        + "_"
        + comparison
        + "_minCount"
        + str(min_count)
        + ".xlsx",
    )
)