def make_test_set(prediction, params):
flutype = params.flutype
if 'toy'!=params.test:
year=params.year
prediction_regions = params.pred.split(',')
test_regions = params.test.split(',')
# define test data sets by choosing start and stop dates
if "oceania" in test_regions:
test_set = {'start':date(year, 3,1), 'stop':date(year, 9,30),
'regions':test_regions, 'sample_size':params.sample_size}
elif 'toy' in params.test:
test_set = {}
test_set['start']= params.gen+(params.valdt - 0.25)*params.dt
test_set['stop'] = params.gen+(params.valdt + 0.25)*params.dt
test_set['regions'] = test_regions
test_set['sample_size'] = 50
else: # assume northern hemisphere
test_set = {'start':date(year, 10,1), 'stop':date(year+1, 3,31),
'regions':test_regions, 'sample_size':params.sample_size}
# load data to test
test_data = PF.flu_alignment(prediction.data.aln_file_name,prediction.data.outgroup,
prediction.data.annotation,cds=prediction.data.cds,
criteria = [[test_set['start'], test_set['stop'],
[reg],test_set['sample_size']] for reg in test_set['regions']],
collapse=params.collapse, build_tree=False)
return test_data, test_set
def make_combined_data(prediction, test_data, otherseqsnames=None, collapse = False):
assert prediction.data.aln.get_alignment_length()==test_data.aln.get_alignment_length(),\
"predict_and_test: prediction and test alignment have different length"
# combined data set needed for plotting
seqname_list = [seq.name for seq in prediction.data.aln] + [seq.name for seq in test_data.aln]
if otherseqsnames:
for seqname in otherseqsnames:
if seqname not in seqname_list:
seqname_list.append(seqname)
combined_data = PF.flu_alignment(test_data.aln_file_name, test_data.outgroup, test_data.annotation,
cds = test_data.cds, seq_names = seqname_list, collapse=collapse)
return combined_data
def predict_params(methods, params):
flutype = params.flutype
if 'toy'!=params.pred:
year=params.year
prediction_regions = params.pred.split(',')
test_regions = params.test.split(',')
pseudo_count = 5
# define prediction and test data sets by choosing start and stop dates
if "oceania" in prediction_regions:
prediction_set={'start': date(year-2, 10,1), 'stop':date(year-1, 9,30),
'regions':prediction_regions, 'sample_size':params.sample_size}
elif "toy" in prediction_regions:
prediction_set={}
prediction_set['start'] = params.gen-0.5*params.dt
prediction_set['stop'] = params.gen+0.5*params.dt
prediction_set['regions'] = prediction_regions
prediction_set['sample_size'] = params.sample_size
else: # assume northern hemisphere
# begin previous year on may 1st, end this year on Feb 28
prediction_set={'start': date(year-1,5,1), 'stop':date(year, 2,28),
'regions':prediction_regions, 'sample_size':params.sample_size}
if 'toy' in params.pred:
aln_fname, annotation, outgroup,cds = get_aln_annotation_outgroup_cds_toy(params)
tbins = None
else:
aln_fname, annotation, outgroup,cds = get_aln_annotation_outgroup_cds(params)
# define 105 day intervals to estimate changing clade frequencies.
# chosen to have 3 intervals between May and Feb
bin_dt = 105
tbins = [ date.fromordinal(prediction_set['stop'].toordinal()-ii*bin_dt) for ii in range(
(prediction_set['stop'].toordinal()-prediction_set['start'].toordinal())//bin_dt,-1,-1)]
prediction_data = PF.flu_alignment(aln_fname,outgroup,annotation,cds=cds, subsample_factor = params.subsample,
criteria = [[prediction_set['start'], prediction_set['stop'],
[reg],prediction_set['sample_size']] for reg in prediction_set['regions']], collapse=params.collapse)
# PREDICT
prediction = PF.flu_ranking(prediction_data, eps_branch_length = params.eps_branch, boost = params.boost,
time_bins = tbins, methods=methods, D=params.diffusion,samp_frac = params.omega,
distance_scale = params.gamma,pseudo_count = pseudo_count)
prediction.predict()
print "prediction done"
return prediction
