def run(self):
"""
2008-11-08
generate combinations of results_id, list_type_id and generate plots one after another
save the plots into database if commit=1
"""
if self.debug:
import pdb
pdb.set_trace()
db = Stock_250kDB.Stock_250kDB(drivername=self.drivername,
username=self.db_user,
password=self.db_passwd,
hostname=self.hostname,
database=self.dbname,
schema=self.schema)
db.setup()
session = db.session
param_obj = PassingData(call_method_id=self.call_method_id, \
analysis_method_id=getattr(self, 'analysis_method_id', None),\
analysis_method_id_ls=getattr(self, 'analysis_method_id_ls', None),\
phenotype_method_id_ls=getattr(self, 'phenotype_method_id_ls', None),\
list_type_id_ls=self.list_type_id_ls, \
results_type=self.results_type)
params_ls = MpiGeneListRankTest.generate_params(param_obj)
ResultsClass, TestResultClass = db.getResultsAndTestResultsClass(
results_type=self.results_type)
if ResultsClass is None or TestResultClass is None:
sys.stderr.write("Invalid results type : %s.\n" % pd.results_type)
sys.exit(3)
for results_id, list_type_id in params_ls:
rm = ResultsClass.get(results_id)
list_type = Stock_250kDB.GeneListType.get(list_type_id)
title = 'result(%s) of %s on %s with %s(%s) list'%\
(results_id, rm.analysis_method.short_name, rm.phenotype_method.short_name, list_type.short_name, list_type.id)
TopSNPTestType_id_ls = self.getTopSNPTestType_id_ls(self.get_closest, self.min_MAF, self.allow_two_sample_overlapping, self.results_type, \
self.test_type_id, self.null_distribution_type_id)
if self.commit:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count() > 0:
row = rows.first()
sys.stderr.write(
'%s already in db (%s of them) with first id=%s.\n' %
(title, rows.count(), row.id))
continue
if not TopSNPTestType_id_ls:
sys.stderr.write(
"No TopSNPTestType matches the input requirements. Exit.\n"
)
sys.exit(3)
TopSNPTestType_id_ls_str = map(str, TopSNPTestType_id_ls)
from_where_clause = "from %s t, %s y where t.type_id=y.id and t.results_id=%s and t.list_type_id=%s and y.id in (%s)"%\
(TestResultClass.table.name, Stock_250kDB.CandidateGeneTopSNPTestRMType.table.name,\
results_id, list_type_id, ','.join(TopSNPTestType_id_ls_str))
no_of_top_snps_info = self.get_no_of_top_snps_info(
db, from_where_clause)
min_distance_info = self.get_min_distance_info(
db, from_where_clause)
rdata = self.get_data_matrix(db, no_of_top_snps_info, min_distance_info, from_where_clause, need_other_values=True, \
null_distribution_type_id=self.null_distribution_type_id)
header = ['no_of_top_snps', ''] + min_distance_info.label_ls
strain_acc_list = no_of_top_snps_info.label_ls
category_list = no_of_top_snps_info.label_ls
if SNPData.isDataMatrixEmpty(rdata.data_matrix):
sys.stderr.write("Nothing fetched from database.\n")
#sys.exit(3)
continue
if self.output_fname:
write_data_matrix(rdata.data_matrix, self.output_fname, header,
strain_acc_list, category_list)
"""
if self.fig_fname:
font = get_font(self.font_path, font_size=self.font_size) #2008-08-01
value2color_func = lambda x: Value2Color.value2HSLcolor(x, rdata.min_value, rdata.max_value)
im_legend = drawContinousLegend(rdata.min_value, rdata.max_value, self.no_of_ticks, value2color_func, font)
#im.save('%s_legend.png'%self.fig_fname_prefix)
im = drawMatrix(rdata.data_matrix, value2color_func, no_of_top_snps_info.label_ls,\
min_distance_info.label_ls, with_grid=1, font=font)
im = combineTwoImages(im, im_legend, font=font)
im.save(self.fig_fname)
"""
if self.commit:
output_fname_prefix = None
else:
title_cp = title
title_cp = title_cp.replace('/', '_')
output_fname_prefix = '%s_%s_type_%s.png' % (os.path.splitext(
self.fig_fname)[0], title_cp, TopSNPTestType_id_ls[0])
if rm.analysis_method_id == 1 or rm.analysis_method_id == 7:
preset_xlim = [0, 8]
preset_xlim = None
else:
preset_xlim = None
return_data = self.plotCurve(rdata,
no_of_top_snps_info,
min_distance_info,
output_fname_prefix,
title=title,
commit=self.commit,
preset_xlim=preset_xlim)
if self.commit and return_data.png_data:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count() > 0:
row = rows.first()
sys.stderr.write(
'%s already in db (%s of them) with first id=%s.\n' %
(title, rows.count(), row.id))
continue
plot = Stock_250kDB.CandidateVsNonRatioPlot(
type_id=TopSNPTestType_id_ls[0],
results_id=results_id,
list_type_id=list_type_id)
plot.png_thumbnail = return_data.png_thumbnail.getvalue()
plot.png_data = return_data.png_data.getvalue()
plot.svg_data = return_data.svg_data.getvalue()
db.session.save(plot)
db.session.flush()
def run(self):
"""
2008-11-08
generate combinations of results_id, list_type_id and generate plots one after another
save the plots into database if commit=1
"""
if self.debug:
import pdb
pdb.set_trace()
db = Stock_250kDB.Stock_250kDB(drivername=self.drivername, username=self.db_user,
password=self.db_passwd, hostname=self.hostname, database=self.dbname, schema=self.schema)
db.setup()
session = db.session
param_obj = PassingData(call_method_id=self.call_method_id, \
analysis_method_id=getattr(self, 'analysis_method_id', None),\
analysis_method_id_ls=getattr(self, 'analysis_method_id_ls', None),\
phenotype_method_id_ls=getattr(self, 'phenotype_method_id_ls', None),\
list_type_id_ls=self.list_type_id_ls, \
results_type=self.results_type)
params_ls = MpiGeneListRankTest.generate_params(param_obj)
for results_id, list_type_id in params_ls:
rm = Stock_250kDB.ResultsMethod.get(results_id)
list_type = Stock_250kDB.GeneListType.get(list_type_id)
title = 'result(%s) of %s on %s with %s(%s) list'%\
(results_id, rm.analysis_method.short_name, rm.phenotype_method.short_name, list_type.short_name, list_type.id)
TopSNPTestType_id_ls = self.getTopSNPTestType_id_ls(self.get_closest, self.min_MAF, self.allow_two_sample_overlapping, self.results_type, \
self.test_type_id, self.null_distribution_type_id)
if self.commit:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count()>0:
row = rows.first()
sys.stderr.write('%s already in db (%s of them) with first id=%s.\n'%(title, rows.count(), row.id))
continue
if not TopSNPTestType_id_ls:
sys.stderr.write("No TopSNPTestType matches the input requirements. Exit.\n")
sys.exit(3)
TopSNPTestType_id_ls_str = map(str, TopSNPTestType_id_ls)
from_where_clause = "from %s t, %s y where t.type_id=y.id and t.results_id=%s and t.list_type_id=%s and y.id in (%s)"%\
(Stock_250kDB.CandidateGeneTopSNPTestRM.table.name, Stock_250kDB.CandidateGeneTopSNPTestRMType.table.name,\
results_id, list_type_id, ','.join(TopSNPTestType_id_ls_str))
no_of_top_snps_info = self.get_no_of_top_snps_info(db, from_where_clause)
min_distance_info = self.get_min_distance_info(db, from_where_clause)
rdata = self.get_data_matrix(db, no_of_top_snps_info, min_distance_info, from_where_clause, need_other_values=True, \
null_distribution_type_id=self.null_distribution_type_id)
header = ['no_of_top_snps', ''] + min_distance_info.label_ls
strain_acc_list = no_of_top_snps_info.label_ls
category_list = no_of_top_snps_info.label_ls
if SNPData.isDataMatrixEmpty(rdata.data_matrix):
sys.stderr.write("Nothing fetched from database.\n")
#sys.exit(3)
continue
if self.output_fname:
write_data_matrix(rdata.data_matrix, self.output_fname, header, strain_acc_list, category_list)
"""
if self.fig_fname:
font = get_font(self.font_path, font_size=self.font_size) #2008-08-01
value2color_func = lambda x: Value2Color.value2HSLcolor(x, rdata.min_value, rdata.max_value)
im_legend = drawContinousLegend(rdata.min_value, rdata.max_value, self.no_of_ticks, value2color_func, font)
#im.save('%s_legend.png'%self.fig_fname_prefix)
im = drawMatrix(rdata.data_matrix, value2color_func, no_of_top_snps_info.label_ls,\
min_distance_info.label_ls, with_grid=1, font=font)
im = combineTwoImages(im, im_legend, font=font)
im.save(self.fig_fname)
"""
if self.commit:
output_fname_prefix = None
else:
title_cp = title
title_cp = title_cp.replace('/', '_')
output_fname_prefix='%s_%s_type_%s.png'%(os.path.splitext(self.fig_fname)[0], title_cp, TopSNPTestType_id_ls[0])
if rm.analysis_method_id ==1 or rm.analysis_method_id==7:
preset_xlim = [0,8]
else:
preset_xlim = None
return_data = self.plotCurve(rdata, no_of_top_snps_info, min_distance_info, output_fname_prefix, title=title, commit=self.commit, preset_xlim=preset_xlim)
if self.commit and return_data.png_data:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count()>0:
row = rows.first()
sys.stderr.write('%s already in db (%s of them) with first id=%s.\n'%(title, rows.count(), row.id))
continue
plot = Stock_250kDB.CandidateVsNonRatioPlot(type_id=TopSNPTestType_id_ls[0], results_id=results_id, list_type_id=list_type_id)
plot.png_thumbnail = return_data.png_thumbnail.getvalue()
plot.png_data = return_data.png_data.getvalue()
plot.svg_data = return_data.svg_data.getvalue()
db.session.save(plot)
db.session.flush()
def run(self):
"""
2008-11-11
"""
if self.debug:
import pdb
pdb.set_trace()
db = Stock_250kDB.Stock_250kDB(drivername=self.drivername, username=self.db_user,
password=self.db_passwd, hostname=self.hostname, database=self.dbname, schema=self.schema)
db.setup()
session = db.session
param_obj = PassingData(call_method_id=self.call_method_id, \
analysis_method_id=getattr(self, 'analysis_method_id', None),\
analysis_method_id_ls=getattr(self, 'analysis_method_id_ls', None),\
phenotype_method_id_ls=getattr(self, 'phenotype_method_id_ls', None),\
list_type_id_ls=self.list_type_id_ls, \
results_type=self.results_type)
params_ls = MpiGeneListRankTest.generate_params(param_obj)
ax = self.init_3d_plot()
result_index = 0
data_to_output_label_ls = []
data_to_output_ls = []
for results_id, list_type_id in params_ls:
rm = Stock_250kDB.ResultsMethod.get(results_id)
list_type = Stock_250kDB.GeneListType.get(list_type_id)
title = '%s on %s_%s (%s)'%\
(rm.analysis_method.short_name, rm.phenotype_method.id, rm.phenotype_method.short_name, results_id)
#a short output label
output_label = '%s_%s (%s)'%\
(rm.phenotype_method.id, rm.phenotype_method.short_name, results_id)
phenotype_label = '%s_%s'%\
(rm.phenotype_method.id, rm.phenotype_method.short_name)
TopSNPTestType_id_ls = self.getTopSNPTestType_id_ls(self.get_closest, self.min_MAF, self.allow_two_sample_overlapping, self.results_type, \
self.test_type_id, self.null_distribution_type_id)
if self.commit:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count()>0:
row = rows.first()
sys.stderr.write('%s already in db (%s of them) with first id=%s.\n'%(title, rows.count(), row.id))
continue
if not TopSNPTestType_id_ls:
sys.stderr.write("No TopSNPTestType matches the input requirements. Exit.\n")
sys.exit(3)
TopSNPTestType_id_ls_str = map(str, TopSNPTestType_id_ls)
from_where_clause = "from %s t, %s y where t.type_id=y.id and t.results_id=%s and t.list_type_id=%s and y.id in (%s)"%\
(Stock_250kDB.CandidateGeneTopSNPTestRM.table.name, Stock_250kDB.CandidateGeneTopSNPTestRMType.table.name,\
results_id, list_type_id, ','.join(TopSNPTestType_id_ls_str))
no_of_top_snps_info = self.get_no_of_top_snps_info(db, from_where_clause)
min_distance_info = self.get_min_distance_info(db, from_where_clause)
rdata = self.get_data_matrix(db, no_of_top_snps_info, min_distance_info, from_where_clause, need_other_values=True, \
null_distribution_type_id=self.null_distribution_type_id)
if rm.analysis_method_id=='6': #For random forest, take log and determine bar length according to score_cutoff_ls
score_cutoff_take_log = True
else:
score_cutoff_take_log = False
return_code = self.plot_one_bar(ax, rdata, no_of_top_snps_info, min_distance_info, self.min_distance, result_index=result_index, data_type=self.data_type, \
output_fname=None, \
need_svg=False, title=phenotype_label, commit=0, preset_xlim =None, score_cutoff_take_log=score_cutoff_take_log)
if return_code:
data_to_output_label_ls.append(output_label)
data_to_output_ls.append(return_code)
result_index += 1
if self.fig_fname:
pylab.savefig(self.fig_fname, dpi=300)
pylab.savefig('%s.svg'%os.path.splitext(self.fig_fname)[0], dpi=300)
#pylab.show()
if self.output_fname and data_to_output_ls:
self.output_data(data_to_output_label_ls, data_to_output_ls, self.min_distance, self.output_fname)
def run(self):
"""
2008-11-11
"""
if self.debug:
import pdb
pdb.set_trace()
db = Stock_250kDB.Stock_250kDB(drivername=self.drivername,
username=self.db_user,
password=self.db_passwd,
hostname=self.hostname,
database=self.dbname,
schema=self.schema)
db.setup()
session = db.session
param_obj = PassingData(call_method_id=self.call_method_id, \
analysis_method_id=getattr(self, 'analysis_method_id', None),\
analysis_method_id_ls=getattr(self, 'analysis_method_id_ls', None),\
phenotype_method_id_ls=getattr(self, 'phenotype_method_id_ls', None),\
list_type_id_ls=self.list_type_id_ls, \
results_type=self.results_type)
params_ls = MpiGeneListRankTest.generate_params(param_obj)
ResultsClass, TestResultClass = Stock_250kDB.Stock_250kDB.getResultsAndTestResultsClass(
results_type=self.results_type)
if ResultsClass is None or TestResultClass is None:
sys.stderr.write("Invalid results type : %s.\n" % pd.results_type)
sys.exit(3)
ax = self.init_3d_plot()
result_index = 0
data_to_output_label_ls = []
data_to_output_ls = []
for results_id, list_type_id in params_ls:
rm = ResultsClass.get(results_id)
list_type = Stock_250kDB.GeneListType.get(list_type_id)
title = '%s on %s_%s (%s)'%\
(rm.analysis_method.short_name, rm.phenotype_method.id, rm.phenotype_method.short_name, results_id)
#a short output label
output_label = '%s_%s (%s)'%\
(rm.phenotype_method.id, rm.phenotype_method.short_name, results_id)
phenotype_label = '%s_%s'%\
(rm.phenotype_method.id, rm.phenotype_method.short_name)
TopSNPTestType_id_ls = self.getTopSNPTestType_id_ls(self.get_closest, self.min_MAF, self.allow_two_sample_overlapping, self.results_type, \
self.test_type_id, self.null_distribution_type_id)
if self.commit:
rows = Stock_250kDB.CandidateVsNonRatioPlot.query.filter_by(type_id=TopSNPTestType_id_ls[0]).\
filter_by(results_id=results_id).filter_by(list_type_id=list_type_id)
if rows.count() > 0:
row = rows.first()
sys.stderr.write(
'%s already in db (%s of them) with first id=%s.\n' %
(title, rows.count(), row.id))
continue
if not TopSNPTestType_id_ls:
sys.stderr.write(
"No TopSNPTestType matches the input requirements. Exit.\n"
)
sys.exit(3)
TopSNPTestType_id_ls_str = map(str, TopSNPTestType_id_ls)
from_where_clause = "from %s t, %s y where t.type_id=y.id and t.results_id=%s and t.list_type_id=%s and y.id in (%s)"%\
(TestResultClass.table.name, Stock_250kDB.CandidateGeneTopSNPTestRMType.table.name,\
results_id, list_type_id, ','.join(TopSNPTestType_id_ls_str))
no_of_top_snps_info = self.get_no_of_top_snps_info(
db, from_where_clause)
min_distance_info = self.get_min_distance_info(
db, from_where_clause)
rdata = self.get_data_matrix(db, no_of_top_snps_info, min_distance_info, from_where_clause, need_other_values=True, \
null_distribution_type_id=self.null_distribution_type_id)
if rm.analysis_method_id == '6': #For random forest, take log and determine bar length according to score_cutoff_ls
score_cutoff_take_log = True
else:
score_cutoff_take_log = False
return_code = self.plot_one_bar(ax, rdata, no_of_top_snps_info, min_distance_info, self.min_distance, result_index=result_index, data_type=self.data_type, \
output_fname=None, \
need_svg=False, title=phenotype_label, commit=0, preset_xlim =None, score_cutoff_take_log=score_cutoff_take_log)
if return_code:
data_to_output_label_ls.append(output_label)
data_to_output_ls.append(return_code)
result_index += 1
ax.set_xlabel('cutoff')
ax.set_ylabel('result')
ax.set_zlabel('data')
if self.fig_fname:
pylab.savefig(self.fig_fname, dpi=300)
pylab.savefig('%s.svg' % os.path.splitext(self.fig_fname)[0],
dpi=300)
#pylab.show()
if self.output_fname and data_to_output_ls:
self.output_data(data_to_output_label_ls, data_to_output_ls,
self.min_distance, self.output_fname)
