Paper figures and scripts -- slight changes to pathoscore output for paper and enduser

paper/annotate.sh

@@ -0,0 +1,59 @@
+#!/bin/bash
+#SBATCH --account=quinlan-kp
+#SBATCH --partition=quinlan-kp
+#SBATCH -o %j-%N.out
+#SBATCH -e %j-%N.err
+#SBATCH --time=48:00:00
+
+#########
+#CLINVAR#
+#########
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/clinvar/clinvar-benign.20170905.vcf.gz --prefix benign
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max  ../truth-sets/GRCh37/clinvar/clinvar-pathogenic-likely_pathogenic.20170905.vcf.gz --prefix pathogenic
+
+# ad gene files
+python ../pathoscore.py annotate pathogenic.vcf.gz --exclude ../gene-sets/GRCh37/ad_genes/ad_gene_complement.bed.gz  --prefix adgene.pathogenic
+python ../pathoscore.py annotate benign.vcf.gz --exclude ../gene-sets/GRCh37/ad_genes/ad_gene_complement.bed.gz  --prefix adgene.benign
+
+# ar gene files
+python ../pathoscore.py annotate pathogenic.vcf.gz --exclude ../gene-sets/GRCh37/ar_genes/ar_gene_complement.bed.gz  --prefix argene.pathogenic
+python ../pathoscore.py annotate benign.vcf.gz --exclude ../gene-sets/GRCh37/ar_genes/ar_gene_complement.bed.gz  --prefix argene.benign
+
+# filtered pathogenics/benigns on ClinVar
+python ../pathoscore.py annotate pathogenic.vcf.gz --exclude ../scripts/gnomad/ExAC.vcf.gz --exclude ../scripts/gnomad/gnomad.vcf.gz --prefix pathogenic.filter
+python ../pathoscore.py annotate benign.vcf.gz --exclude ../scripts/gnomad/ExAC.vcf.gz --exclude ../scripts/gnomad/gnomad.vcf.gz --prefix benign.filter
+
+#########
+#SAMOCHA#
+#########
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/samocha/samocha.benign.vcf.gz --prefix control
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/samocha/samocha.pathogenic.vcf.gz --prefix neurodev
+
+# filtered pathogenics on samocha
+python ../pathoscore.py annotate neurodev.vcf.gz --exclude ../scripts/gnomad/ExAC.vcf.gz --exclude ../scripts/gnomad/gnomad.vcf.gz --prefix neurodev.filter
+
+#######
+#HOMSY#
+#######
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/homsy/homsy.benign.vcf.gz --prefix hombenign
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/homsy/homsy.pathogenic.vcf.gz --prefix hompathogenic
+
+# filtered pathogenics on homsy
+python ../pathoscore.py annotate hompathogenic.vcf.gz --exclude ../scripts/gnomad/ExAC.vcf.gz --exclude ../scripts/gnomad/gnomad.vcf.gz --prefix hompathogenic.filter
+
+########
+#WIGLER#
+########
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/wigler-autism/wigler-autism.benign.vcf.gz --prefix wigbenign
+
+python ../pathoscore.py annotate --scores ../score-sets/GRCh37/CCR/ccrs.autosomes.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CCR/ccrs.xchrom.v2.20180420.bed.gz:CCR:4:max --scores ../score-sets/GRCh37/CADD/whole_genome_SNVs.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/CADD/InDels.tsv.gz:CADD:6:max --scores ../score-sets/GRCh37/DANN/DANN_whole_genome_SNVs.tsv.bgz:DANN:5:max --scores ../score-sets/GRCh37/GERP/gerp_rs.txt.gz:GERP:3:max --scores ../score-sets/GRCh37/MCAP/mcap.txt.gz:MCAP:5:max --scores ../score-sets/GRCh37/MPC/mpc.txt.gz:MPC:5:max --scores ../score-sets/GRCh37/MTR/mtrflatfile_1.0.txt.gz:MTR:11:max --scores ../score-sets/GRCh37/REVEL/revel.txt.gz:REVEL:7:max --scores ../score-sets/GRCh37/RVIS/rvis.bed.gz:RVIS:5:max --scores ../score-sets/GRCh37/VVP/VVP_scores_allChr.txt.gz:VVP:5:max --scores ../score-sets/GRCh37/aloft/aloft.txt.gz:aloft_het,aloft_lof,aloft_rec:5,6,7:max,max,max --scores ../score-sets/GRCh37/fathmm/fathmm/fathmm.txt.gz:fathmm_non,fathmm_coding:5,6:max,max --scores ../score-sets/GRCh37/fitcons/fitcons/fitcons.bed.gz:fitCons:4:max --scores ../score-sets/GRCh37/metasvm/metasvm.txt.gz:MetaSVM:5:max --scores ../score-sets/GRCh37/mis_Z/missensez.bed.gz:missense_z:5:max --scores ../score-sets/GRCh37/pLI/pLI.bed.gz:pLI:5:max --scores ../score-sets/GRCh37/phastCons/phastcons_vertebrate.txt.gz:phastCons:4:max --scores ../score-sets/GRCh37/polyphen2/polyphen2/polyphen2.txt.gz:polyphen2_hvar:6:max --scores ../score-sets/GRCh37/sift/sift.txt.gz:SIFT:12:max ../truth-sets/GRCh37/wigler-autism/wigler-autism.pathogenic.vcf.gz --prefix wigpathogenic
+
+# filtered pathogenics on wigler
+python ../pathoscore.py annotate wigpathogenic.vcf.gz --exclude ../scripts/gnomad/ExAC.vcf.gz --exclude ../scripts/gnomad/gnomad.vcf.gz --prefix wigpathogenic.filter

paper/correlation.py

@@ -0,0 +1,58 @@
+from __future__ import print_function
+metrics = ["CCR", "CADD", "DANN", "GERP", "MCAP", "MPC", "MTR", "REVEL", "RVIS", "VVP", "aloft_het", "aloft_lof", "aloft_rec", "fathmm_non", "fathmm_coding", "fitCons", "MetaSVM", "missense_z", "pLI", "phastCons", "polyphen2_hvar", "SIFT", "mis_badness", "Grantham", "BLOSUM"]
+from cyvcf2 import VCF
+import pandas as pd
+from argparse import ArgumentParser
+import matplotlib
+matplotlib.use('Agg')
+import matplotlib.backends.backend_pdf
+from matplotlib import pyplot as plt
+from matplotlib import ticker
+#import seaborn as sns
+#sns.set_style('white')
+import sys
+if sys.version_info.major > 2:
+    xrange = range
+
+def key_val(arg):
+    return arg.split(",")
+parser = ArgumentParser(description = "create correlation matrix for metrics across different truth sets")
+parser.add_argument("-f", "--file", help = "input vcfs", type=key_val, nargs = "+")
+parser.add_argument("-n", "--name", help = "evaluation set names/titles", nargs = "+")
+parser.add_argument("-o", "--output", help = "output file")
+args = parser.parse_args()
+
+corr = {}
+for var, name in zip(args.file, args.name):
+    patho, benign = var
+    vcfpatho = VCF(patho)
+    vcfbenign = VCF(benign)
+    p = pd.DataFrame([variant.INFO.get(key) for key in metrics] for variant in vcfpatho)
+    b = pd.DataFrame([variant.INFO.get(key) for key in metrics] for variant in vcfbenign)
+    p.columns = metrics
+    b.columns = metrics
+    d = p.append(b)
+    corr[name] = d.corr(method = 'kendall')
+    print(corr)
+print(d)
+
+plt.rcParams["figure.figsize"]=(5,5)
+matplotlib.rcParams['pdf.fonttype'] = 42
+matplotlib.rcParams['font.family'] = 'sans-serif'
+matplotlib.rcParams['font.sans-serif'] = ['Arial']
+print(len(args.file))
+#fig, ax = plt.subplots(len(args.file), 1)
+for i, name in enumerate(args.name):
+    fig, ax = plt.subplots(1)
+    g = ax.matshow(corr[name], interpolation = 'nearest', cmap = plt.cm.Blues)
+    fig.colorbar(g, ax=ax)
+    ax.xaxis.set_major_locator(ticker.LinearLocator(numticks = len(metrics)))
+    ax.yaxis.set_major_locator(ticker.LinearLocator(numticks = len(metrics)))
+    ax.set_xticklabels(metrics)
+    ax.set_yticklabels(metrics)
+    plt.setp(ax.get_xticklabels(), fontsize = 10, rotation = 'vertical')
+    plt.title(name, y=1.35)
+pdf = matplotlib.backends.backend_pdf.PdfPages(args.output)
+for fig in xrange(1, plt.gcf().number+1): ## will open an empty extra figure :(
+    pdf.savefig(fig, bbox_inches = 'tight')
+pdf.close()

paper/correlation.sh

@@ -0,0 +1 @@
+python correlation.py -f pathogenic.vcf.gz,benign.vcf.gz -n ClinVar -o $HOME/public_html/pathoscorepaper/correlation.pdf

paper/create.sh

@@ -0,0 +1,136 @@
+#create score sets
+
+cd ../score-sets/GRCh37/CADD
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/CCR
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/DANN
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/GERP
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/MCAP
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/MPC
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/MTR
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/REVEL
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/RVIS
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/VVP
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/aloft
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/fathmm
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/fitcons
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/metasvm
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/mis_Z
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/pLI
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/phastCons
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/phylop
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/polyphen2  
+bash make.sh
+cd -
+
+cd ../score-sets/GRCh37/sift
+bash make.sh
+cd -
+
+# make truth sets
+
+cd ../truth-sets/GRCh37/clinvar
+bash make.sh
+cd -
+
+cd ../truth-sets/GRCh37/homsy
+bash make.sh
+cd -
+
+cd ../truth-sets/GRCh37/samocha
+bash make.sh
+cd -
+
+cd ../truth-sets/GRCh37/wellderly
+bash make.sh
+cd -
+
+cd ../truth-sets/GRCh37/wigler-autism
+bash make.sh
+cd -
+
+# make gene sets
+
+cd ../gene-sets/GRCh37/ad_genes/
+bash make.sh
+cd -
+
+cd ../gene-sets/GRCh37/hi_genes/
+bash make.sh
+cd -
+
+cd ../gene-sets/GRCh37/ar_genes/
+bash make.sh
+cd -
+
+cd ../gene-sets/GRCh37/cancer_genes/
+bash make.sh
+cd -
+
+cd ../gene-sets/GRCh37/heart_genes/
+bash make.sh
+cd -
+
+cd ../gene-sets/GRCh37/eiee_genes/
+bash make.sh
+cd -
+
+# make population sets
+
+cd ../scripts/gnomad
+bash makeexac.sh
+bash makegnomad.sh
+cd -

paper/cubarplot.py

@@ -0,0 +1,18 @@
+from __future__ import print_function
+import numpy as np
+import matplotlib
+matplotlib.use('Agg')
+from matplotlib import pyplot as plt
+import sys
+import pandas as pd
+
+plt.rcParams["figure.figsize"]=(14,5)
+with open(sys.argv[1], 'r') as clinicaltable:
+    df = pd.read_csv(clinicaltable, sep='\t', index_col=0)
+    df = df.T
+
+ax = df.plot.bar(rot=0,legend=False,width=.8,align='center', fontsize=18)
+ax.legend(loc='center left', bbox_to_anchor=(1, 0.5), fontsize=18)
+ax.set_ylabel("Clinical Utility", fontsize=18)
+plt.tight_layout()
+plt.savefig("".join(sys.argv[1].split(".cu.tsv",)[0])+"cubarplot.pdf", bbox_inches="tight")

paper/cubarplots.sh

@@ -0,0 +1,2 @@
+python cubarplot.py $HOME/public_html/pathoscorepaper/clinvar.cu.tsv
+python cubarplot.py $HOME/public_html/pathoscorepaper/samocha.cu.tsv