def add_scores(self, df):
seqs = list(df['Sequence'])
miss_seqs = list(df['Missing'])
ns = NormScore()
lc_raw = tools_lc.calc_lc_motifs(seqs, self.k, self.alph_lca,
self.thresh_lce)
lc_norms = ns.lc_norm_score(seqs)
lengths = tools_fasta.get_lengths(seqs)
miss_count = self.get_missing(miss_seqs)
df['Length'] = lengths
df['Miss Count'] = miss_count
df['LC Norm'] = lc_norms
df['LC Raw'] = lc_raw
return df
def write_tsv(self):
"""Write a tsv file that is score, nomiss_score, length"""
df = pd.read_csv(self.norm_fpi, sep='\t', index_col=0)
seqs = df['Sequence']
miss_seqs = df['Missing']
lens = [len(seq) for seq in seqs]
raw_scores = tools_lc.calc_lc_motifs(seqs, self.k, self.lca, self.lce)
nomiss_scores = tools_lc.calc_lc_motifs_nomiss(seqs, miss_seqs, self.k,
self.lca, self.lce)
df_dict = {
'score': raw_scores,
'nomiss_score': nomiss_scores,
'Length': lens
}
cols = ['score', 'nomiss_score', 'Length']
df = pd.DataFrame(df_dict, columns=cols)
df.to_csv(self.fpo, sep='\t')
def get_raw_scores(self, lc_label, k, lca, lce):
if lc_label == 'LCA || LCE':
scores = tools_lc.calc_lc_motifs(self.seqs, k, lca, lce)
elif lc_label == 'LCA & LCE':
scores = []
for seq in self.seqs:
scores.append(tools_lc.count_lca_and_lce(seq, k, lca, lce))
elif lc_label == 'LCA & ~LCE':
scores = []
for seq in self.seqs:
scores.append(tools_lc.count_lca_not_lce(seq, k, lca, lce))
elif lc_label == '~LCA & LCE':
scores = []
for seq in self.seqs:
scores.append(tools_lc.count_not_lca_lce(seq, k, lca, lce))
else:
raise Exception('Unexpected logical expression')
return scores
def lc_norm_score(self, seqs):
lens = [len(seq) for seq in seqs]
scores = tools_lc.calc_lc_motifs(seqs, self.k, self.lca, self.lce)
lc_norm = self.norm_function(self.lc_m, self.lc_b, scores, lens)
return lc_norm
def count_lca_or_lce(self, seqs):
lc_motifs = tools_lc.calc_lc_motifs(seqs, self.k, self.lca, self.lce)
return lc_motifs
def mb_lc(self, k, lca, lce):
"""LCA || LCE"""
scores = tools_lc.calc_lc_motifs(self.seqs, k, lca, lce)
lr = linregress(self.lens, scores)
return lr