def score_network(kinase_file, out_file):
# This takes network files and scores all edges according to pssm
ktable = pssms.read_kin_sub_data("data/psiteplus-kinase-substrates.tsv")
aa_freqs = pssms.read_aa_freqs("data/aa-freqs.tsv")
psites = read_phosphosites("data/psiteplus-phosphosites.tsv")
kinases = read_kinase_file(kinase_file)
phosfun_st, med_phosfun_st = read_phosfun_file("data/phosfun-ST.tsv")
phosfun_y, med_phosfun_y = read_phosfun_file("data/phosfun-Y.tsv")
signs = read_sign_file("out/reg-site-bart-sign-preds.tsv")
reg_sites = read_reg_sites_file("data/psiteplus-reg-sites.tsv")
kin_types = read_pfam_file("data/human-pfam.tsv", kinases)
kinase_fams, family_kins = read_kinase_families("data/kinase-families.tsv")
sdr_assigns = read_sdr_assignments("data/sdr-pssm-assignments.tsv")
family_pssms = pssms.build_family_pssms(family_kins, ktable, aa_freqs)
kinase_pssms = pssms.build_pssms(kinases, kinase_fams, family_pssms, ktable,
aa_freqs)
kinase_pssms = assign_pssms_by_sdr(kinases, kinase_pssms, sdr_assigns)
scores = score_kin_pairs(kinases, kinase_pssms, psites, family_pssms)
# scores = norm_scores
with open(out_file, "w") as v:
v.write("\t".join(["node1", "node2", "kinase.type", "sub.kinase.type",
"max.pssm.score", "signed.dcg", "site.sign.score",
"signed.func.score"])+"\n")
for pair in scores:
if pair[0] not in kin_types:
if pair[0] == "MTOR":
kin_type = "ST"
else:
kin_type = "NA"
else:
kin_type = kin_types[pair[0]]
if pair[1] not in kin_types:
if pair[1] == "MTOR":
sub_type = "ST"
else:
sub_type = "NA"
else:
sub_type = kin_types[pair[1]]
if not scores[pair]:
v.write("\t".join([pair[0], pair[1], kin_type, sub_type, "NA",
"NA", "NA", "NA"])+"\n")
continue
own_pssm = str(kinase_pssms[pair[0]][1]).upper()
residues = set([res for pos, res, score in scores[pair]])
max_score = max([score for pos, res, score in scores[pair]])
if residues == set(['Y']):
dcg, site_score, signed_func_score = signed_regulation_score(
pair, scores[pair], phosfun_y, med_phosfun_y, reg_sites, signs)
else:
dcg, site_score, signed_func_score = signed_regulation_score(
pair, scores[pair], phosfun_st, med_phosfun_st, reg_sites, signs)
v.write("\t".join([pair[0], pair[1], kin_type, sub_type, str(max_score),
str(dcg), str(site_score), str(signed_func_score)])+"\n")
def calc_distr(kin_act_file):
ktable = pssms.read_kin_sub_data("data/reduced_kinase_table.tsv")
aa_freqs = pssms.read_aa_freqs("data/aa-freqs.tsv")
proteome = pssms.read_proteome()
out_file_base = os.path.splitext(os.path.basename(kin_act_file))[0]
dist_out_file = "out/" + out_file_base + "-pssm-dists.tsv"
with open(dist_out_file, "w") as v:
for kinase in ktable:
kin_motif_seqs = ktable[kinase]
motif_seqs = [seq for (substrate, seq) in kin_motif_seqs]
pssm = pssms.calc_pssm(motif_seqs, aa_freqs)
sub_motif_seqs = get_random_sty_motifs(proteome, kin_motif_seqs,
1000)
dist = calc_dist(sub_motif_seqs, pssm)
if not dist:
continue
line = [kinase]
line.extend([str(x) for x in dist])
v.write("\t".join(line) + "\n")
def pssm_prop(seqs):
if len(seqs) == 0:
return (np.zeros((20, 2)))
#this function returns simple proportinal pssm
pssm = np.zeros((20, len(seqs[0])))
for seq in seqs:
for i, aa in enumerate(seq):
if aa not in AMINO_ACIDS:
continue
pssm[AMINO_ACIDS[aa], i] += 1 / len(seqs)
return (pssm)
if __name__ == "__main__":
ktable = pssms.read_kin_sub_data("data/psiteplus-kinase-substrates.tsv")
aa_freqs = pssms.read_aa_freqs("data/aa-freqs.tsv")
psites = read_phosphosites("data/pride-phosphosites.tsv")
kinases = read_kinase_file("data/human-kinome.txt")
# phosfun, med_phosfun = read_phosfun_file("data/phosfun-alt.tsv")
phosfun_st, med_phosfun_st = read_phosfun_file("data/phosfun-ST.tsv")
phosfun_y, med_phosfun_y = read_phosfun_file("data/phosfun-Y.tsv")
reg_sites = read_reg_sites_file("data/psiteplus-reg-sites.tsv")
entropies = {}
pssm_div = {}
for kinase in kinases:
if kinase not in ktable:
continue
kin_motif_seqs = []
for substrate, pos, res, seq in ktable[kinase]:
def score_network(kinase_file, out_file):
# This takes network files and scores all edges according to pssm
ktable = pssms.read_kin_sub_data("data/psiteplus-kinase-substrates.tsv")
aa_freqs = pssms.read_aa_freqs("data/aa-freqs.tsv")
psites = read_phosphosites("data/psiteplus-phosphosites.tsv")
kinases = read_kinase_file(kinase_file)
# phosfun, med_phosfun = read_phosfun_file("data/phosfun-alt.tsv")
phosfun_st, med_phosfun_st = read_phosfun_file("data/phosfun-ST.tsv")
phosfun_y, med_phosfun_y = read_phosfun_file("data/phosfun-Y.tsv")
reg_sites = read_reg_sites_file("data/psiteplus-reg-sites.tsv")
kin_types = read_pfam_file("data/human-pfam.tsv", kinases)
kinase_fams, family_kins = read_kinase_families("data/kinase-families.tsv")
sdr_assigns = read_sdr_assignments("data/sdr-pssm-assignments.tsv")
family_pssms = pssms.build_family_pssms(family_kins, ktable, aa_freqs)
kinase_pssms = pssms.build_pssms(kinases, kinase_fams, family_pssms,
ktable, aa_freqs)
kinase_pssms = assign_pssms_by_sdr(kinases, kinase_pssms, sdr_assigns)
scores = score_kin_pairs(kinases, kinase_pssms, psites, family_pssms)
norm_scores = normalize_scores(scores)
# scores = norm_scores
kinases_sorted = list(kinases)
kinases_sorted.sort()
pairs = itertools.product(kinases_sorted, repeat=2)
with open(out_file, "w") as v:
v.write("\t".join([
"node1", "node2", "kinase.type", "sub.kinase.type", "pssm.source",
"max.pssm.score", "dcg", "max.func.score", "max.pssm.score.res",
"max.func.score.res", "all.pssm.scores", "all.func.scores",
"all.sites"
]) + "\n")
for pair in pairs:
if pair[0] == pair[1]:
continue
if pair[0] not in kin_types:
if pair[0] == "MTOR":
kin_type = "ST"
else:
kin_type = "NA"
else:
kin_type = kin_types[pair[0]]
if pair[1] not in kin_types:
if pair[1] == "MTOR":
sub_type = "ST"
else:
sub_type = "NA"
else:
sub_type = kin_types[pair[1]]
if pair[0] in kinase_pssms:
source = kinase_pssms[pair[0]][1]
else:
source = "NA"
if not scores[pair]:
v.write("\t".join([
pair[0], pair[1], kin_type, sub_type, source, "NA", "NA",
"NA", "NA", "NA", "NA", "NA", "NA"
]) + "\n")
continue
own_pssm = str(kinase_pssms[pair[0]][1]).upper()
residues = set([res for pos, res, score in scores[pair]])
pssm_scores = ([score for pos, res, score in scores[pair]])
sites = ([
"{0}{1}".format(res, pos) for pos, res, score in scores[pair]
])
max_pssm_score = max(pssm_scores)
max_pssm_score_i = pssm_scores.index(max_pssm_score)
max_pssm_score_site = sites[max_pssm_score_i]
all_pssm_scores = ",".join([str(score) for score in pssm_scores])
all_sites = ",".join(sites)
if residues == set(['Y']):
(dcg, max_func_score, max_func_score_site,
all_func_scores) = regulation_score(pair, scores[pair],
phosfun_y, med_phosfun_y,
reg_sites)
else:
(dcg, max_func_score, max_func_score_site,
all_func_scores) = regulation_score(pair, scores[pair],
phosfun_st,
med_phosfun_st, reg_sites)
v.write("\t".join([
pair[0], pair[1], kin_type, sub_type, source,
str(max_pssm_score),
str(dcg),
str(max_func_score), max_pssm_score_site, max_func_score_site,
all_pssm_scores, all_func_scores, all_sites
]) + "\n")
"C": 11,
"M": 12,
"N": 13,
"Q": 14,
"K": 15,
"R": 16,
"H": 17,
"D": 18,
"E": 19
}
if __name__ == "__main__":
disc = "out/discarded-kinases.tsv"
kept = "out/kept-kinases.tsv"
d = pssms.read_kin_sub_data(disc)
k = pssms.read_kin_sub_data(kept)
daa = {}
kaa = {}
for kinase in d:
noas = [0] * 15
for i in range(len(d[kinase][0][3])):
colaa = set()
if i == 7:
continue
for j in range(len(d[kinase])):
if d[kinase][j][3][i] not in colaa and d[kinase][j][3][
i] in AMINO_ACIDS: