def make_linreg(expt_dir,
name,
gene_seq_fname,
gene_len_fname,
tr_codons_fname,
te_codons_fname,
outputs_fname,
rel_cod_idxs=False,
rel_nt_idxs=False,
rel_struc_idxs=False,
struc_fname=False):
linreg.linreg_init(expt_dir,
name,
gene_seq_fname,
gene_len_fname,
tr_codons_fname,
te_codons_fname,
outputs_fname,
rel_cod_idxs=rel_cod_idxs,
rel_nt_idxs=rel_nt_idxs)
len_dict = proc.get_len_dict(gene_len_fname)
cds_dict = proc.get_cds_dict(gene_seq_fname, len_dict)
if struc_fname:
struc_dict = proc.get_struc_dict(struc_fname)
else:
struc_dict = False
X_tr, X_te, y_tr, y_te = proc.get_data_matrices(
cds_dict,
tr_codons_fname,
te_codons_fname,
outputs_fname,
rel_cod_idxs=rel_cod_idxs,
rel_nt_idxs=rel_nt_idxs,
rel_struc_idxs=rel_struc_idxs,
struc_dict=struc_dict)
wts, y_tr_hat, y_te_hat = linreg.train_and_predict(X_tr, X_te, y_tr, y_te)
linreg_fname = "{0}/linreg/{1}".format(expt_dir, name)
proc.pickle_obj(wts, "{0}/wts.pkl".format(linreg_fname))
proc.pickle_obj(y_tr, "{0}/y_tr.pkl".format(linreg_fname))
proc.pickle_obj(y_tr_hat, "{0}/y_tr_hat.pkl".format(linreg_fname))
proc.pickle_obj(y_te, "{0}/y_te.pkl".format(linreg_fname))
proc.pickle_obj(y_te_hat, "{0}/y_te_hat.pkl".format(linreg_fname))
return wts, y_tr_hat, y_te_hat, y_tr, y_te
# Init vars
rel_cod_idxs = range(-5, 5)
rel_nt_idxs = range(-15, 15)
rel_struc_idxs = range(-17, -14)
nn_dir = nn_parent_dir + "/" + model_name +\
"_rep{0}".format(model_rep)
cod_trunc_5p = 20
cod_trunc_3p = 20
min_tot_cts = 200
min_cod_w_cts = 100
# Load transcriptome dicts
len_dict = proc.get_len_dict(gene_len_fname)
cds_dict = proc.get_cds_dict(gene_seq_fname, len_dict)
struc_dict = proc.get_struc_dict(struc_fname)
# Compute cts_by_codon and outputs
cts_by_codon = proc.load_cts_by_codon(cts_by_codon_fname)
outputs = proc.load_outputs(outputs_fname)
paralog_groups = proc.load_paralog_groups(paralog_groups_fname)
# Get gene list to compute performance metrics
gene_set = cts_by_codon.keys()
# Filter genes that are shorter than truncation regions
gene_set = filter(lambda gene: len(cts_by_codon[gene]) > (cod_trunc_5p +
cod_trunc_3p), gene_set)
# Filter genes that don't have enough counts to meet cutoffs
gene_set = filter(lambda gene: proc.has_enough_cts( \
cts_by_codon[gene][cod_trunc_5p:-cod_trunc_3p], min_tot_cts, \
def process_sam_file(expt_dir,
sam_fname,
gene_seq_fname,
gene_len_fname,
shift_dict,
cod_trunc_5p,
cod_trunc_3p,
min_fp_size,
max_fp_size,
num_tr_genes,
num_te_genes,
min_cts_per_gene,
min_cod_w_data,
raw_psct=0,
paralog_groups_fname=False,
overwrite=False,
folds=False):
"""
Processes an RP sam file for an experiment.
Makes in expt_dir/process:
cts_by_codon file (sum sam map wts per codon)
outputs file (scaled cts_by_codon, each gene mean centered at 1)
te_bounds file (first and last codon idxs per test set gene)
te_data_table file (data table for test set codons)
tr_bounds file (first and last codon idxs per training set gene)
tr_data_table file (data table for training set codons)
Args:
expt_dir (str) - name of experiment directory
sam_fname (str) - name of input sam file
gene_seq_fname (str) - name of transcriptome fasta file
gene_len_fname (str) - name of gene lengths file
shift_dict (dict):
{fp_size (int):
{frame (int): shift (int, or False) for frame in range(2)}
for fp_size in range(min_fp_size, max_fp_size + 1) }
cod_trunc_5p (int): number of codons to exclude at start of each CDS
cod_trunc_3p (int): number of codons to exclude at end of each CDS
min_fp_size (int): minimum size footprint to accept in sam file
max_fp_size (int): maximum size footprint to accept in sam file
num_tr_genes (int): number of genes to sort into training set
num_te_genes (int): number of genes to sort into test set
min_cts_per_gene (int):
cutoff for total cts on gene to include gene in tr/te sets
min_cod_w_data (int):
cutoff for codons with data to include gene in tr/te sets
raw_psct (float): psct to add to raw cts_by_codon values
paralog_groups_fname (str):
file containing one group of mutually paralogous genes on each line
results in filtering out all but one paralog (NOTE which one?)
overwrite (bool): flag to overwrite processed files, default False
folds (bool):
might implement this later to divide genes into folds
rather than explicit training/test sets
Returns:
void, makes files listed above
"""
# Load CDS dict, len dict
len_dict = proc.get_len_dict(gene_len_fname)
cds_dict = proc.get_cds_dict(gene_seq_fname, len_dict)
# Process and write cts by codon file
cts_by_codon = proc.get_cts_by_codon(sam_fname, cds_dict, len_dict,
shift_dict, min_fp_size, max_fp_size)
cts_by_codon_fname = expt_dir + \
"/process/cts_by_codon.size.{0}.{1}.txt".format(
min_fp_size, max_fp_size)
if not os.path.isfile(cts_by_codon_fname):
print "making file " + cts_by_codon_fname
proc.write_cts_by_codon(cts_by_codon_fname, cts_by_codon)
else:
print "file " + cts_by_codon_fname + " already exists"
#Process and write outputs file
outputs = proc.get_outputs(cts_by_codon,
cod_trunc_5p,
cod_trunc_3p,
raw_psct=raw_psct)
outputs_fname = expt_dir + \
"/process/outputs.size.{0}.{1}.txt".format(min_fp_size, max_fp_size)
if raw_psct:
outputs_fname = outputs_fname[:-4] +\
".raw_psct.{0}.txt".format(raw_psct)
if not os.path.isfile(outputs_fname):
print "making file " + outputs_fname
proc.write_outputs(outputs_fname, outputs)
else:
print "file " + outputs_fname + " already exists"
#Make training and test set codon files
tr_set_fname = expt_dir + "/process/tr_set_bounds.size." + \
"{0}.{1}.trunc.{2}.{3}.min_cts.{4}.min_cod.{5}.top.{6}.txt".format(
min_fp_size, max_fp_size, cod_trunc_5p, cod_trunc_3p,
min_cts_per_gene, min_cod_w_data, num_tr_genes + num_te_genes)
te_set_fname = expt_dir + "/process/te_set_bounds.size." + \
"{0}.{1}.trunc.{2}.{3}.min_cts.{4}.min_cod.{5}.top.{6}.txt".format(
min_fp_size, max_fp_size, cod_trunc_5p, cod_trunc_3p,
min_cts_per_gene, min_cod_w_data, num_tr_genes + num_te_genes)
print "making file " + tr_set_fname
print "making file " + te_set_fname
proc.make_codon_set_files(cds_dict,
cts_by_codon_fname,
outputs_fname,
tr_set_fname,
te_set_fname,
num_tr_genes,
num_te_genes,
cod_trunc_5p,
cod_trunc_3p,
min_cts_per_gene,
min_cod_w_data,
paralog_groups_fname=paralog_groups_fname,
overwrite=overwrite)