def importGroups(fn):
try: group_db=collections.OrderedDict()
except Exception:
try:
import ordereddict
group_db=ordereddict.OrderedDict()
except Exception: group_db={}
for line in open(fn,'rU').xreadlines():
data = ExpressionBuilder.cleanUpLine(line)
sample_filename,group_number,group_name = string.split(data,'\t')
try: group_db[group_name].append(sample_filename)
except Exception: group_db[group_name] = [sample_filename]
return group_db
def performGroupNormalization(filename,export_dir,platform):
expressionDataFormat,increment,convertNonLogToLog = ExpressionBuilder.checkExpressionFileFormat(filename)
groups_dir = string.replace(export_dir,'exp.','batch.')
fn=unique.filepath(filename); row_number=0; exp_db={}; relative_headers_exported = False
group_db = importGroups(groups_dir)
export_data = export.ExportFile(export_dir)
for line in open(fn,'rU').xreadlines():
data = ExpressionBuilder.cleanUpLine(line)
t = string.split(data,'\t')
if data[0]=='#' and row_number==0: row_number = 0
elif row_number==0:
sample_list = t[1:]
new_sample_list = []
for group in group_db:
group_samples = group_db[group]
try:
sample_index_list = map(lambda x: sample_list.index(x), group_samples)
group_db[group] = sample_index_list
new_sample_list+=group_samples
except Exception:
missing=[]
for x in sample_list:
if x not in t[1:]: missing.append(x)
print 'missing:',missing
print t
print sample_list
print filename, groups_dir
print 'Unknown Error!!! Skipping cluster input file build (check column and row formats for conflicts)'; forceExit
title = string.join([t[0]]+new_sample_list,'\t')+'\n' ### output the new sample order (group file order)
export_data.write(title)
row_number=1
else:
gene = t[0]
if expressionDataFormat == 'non-log' and (convertNonLogToLog or platform == 'RNASeq'):
### Convert to log2 RPKM values - or counts
try: all_values = map(lambda x: math.log(float(x)+increment,2), t[1:])
except Exception:
all_values = ExpressionBuilder.logTransformWithNAs(t[1:],increment)
else:
try: all_values = map(float,t[1:])
except Exception:
all_values = ExpressionBuilder.logTransformWithNAs(t[1:],increment)
row_number+=1 ### Keep track of the first gene as to write out column headers for the relative outputs
gene_log_folds = []
for group in group_db:
sample_index_list = group_db[group]
### Calculate log-fold values relative to the mean of all sample expression values
try: values = map(lambda x: all_values[x], sample_index_list) ### simple and fast way to reorganize the samples
except Exception:
print len(values), sample_index_list;kill
try: avg = statistics.avg(values)
except Exception:
values2=[]
for v in values:
try: values2.append(float(v))
except Exception: pass
values = values2
try: avg = statistics.avg(values)
except Exception:
if len(values)>0: avg = values[0]
else: avg = 0
try: log_folds = map(lambda x: (x-avg), values)
except Exception:
log_folds=[]
for x in values:
try: log_folds.append(x-avg)
except Exception: log_folds.append('')
gene_log_folds+=log_folds
gene_log_folds = map(lambda x: str(x),gene_log_folds)
export_data.write(string.join([gene]+gene_log_folds,'\t')+'\n')
export_data.close()
