def test1():
import time
cal_id = 3
dep_id = 105
t_start = 1407452400
t_end = 1407455985 - (59 * 60)
db_con = util.get_db('reader')
sv = signal1.SteeringVectors(db_con, cal_id)
signal = signal1.Signal(db_con, dep_id, t_start, t_end)
sites = util.get_sites(db_con)
(center, zone) = util.get_center(db_con)
assert zone == util.get_utm_zone(db_con)
start = time.time()
pos = PositionEstimator(dep_id, sites, center, signal, sv,
method=signal1.Signal.MLE)
print "Finished in {0:.2f} seconds.".format(time.time() - start)
print compute_conf(pos.p, pos.num_sites, sites, pos.splines)
var_name='otu',
value_name='abun')
dflong = dflong.merge(meta[['site', 'subject_id']],
left_on='sample',
right_index=True)
# Remove any samples corresponding to second time point
# Fundo samples end in F2 (corresponding to patient X-F1)
# gastric/throat time points end in GI/GF or TI/TF
exclude = ['2', 'F', 'sick', 'F2T']
for s in exclude:
dflong = dflong[~dflong['sample'].str.endswith(s)]
# And remove any lung transplant samples
dflong = dflong[~dflong['sample'].str.startswith('05')]
## Correlations
sites = util.get_sites()
res = []
## Set up the inputs to parallel code
fxn_data = []
print('Setting up input data...')
for n in range(args.nshuffle):
print(n)
# Shuffle patient IDs within site-OTU combination
dflong['subject_id'] = dflong\
.groupby(['site', 'otu'])['subject_id']\
.transform(np.random.permutation)
# This is very slow. I could probably make it faster somehow?
for o, otudf in dflong.groupby('otu'):
# For each pairwise site combo:
for site1 in sites: