def ttest_stats(group1, group2, dst, cond):
"""t-Test on differences between the individual voxels in group1 and group2, group is a array of voxelizations"""
#make sure destination dir exists
if not os.path.exists(dst): os.mkdir(dst)
g1 = stat.readDataGroup(group1)
g2 = stat.readDataGroup(group2)
tvals, pvals = stats.ttest_ind(g1, g2, axis=0, equal_var=False)
np.save(os.path.join(dst, "tvals_%s.npy" % cond), tvals)
np.save(os.path.join(dst, "pvals_%s.npy" % cond), pvals)
print(
"\nsaved t-statistics and p-values corresponding to registered voxels as numpy arrays \n\n."
)
pvals_flat = np.ravel(pvals)
tvals_flat = np.ravel(tvals)
pvals_flat = pvals_flat[~np.isnan(pvals_flat)]
tvals_flat = tvals_flat[~np.isnan(tvals_flat)]
pvals_flat.shape
df = pd.DataFrame()
df['tvalues'] = tvals_flat
df['pvalues'] = pvals_flat
df.to_csv(os.path.join(dst, "tstats_no_nans_%s.csv" % cond))
return
xx +
"/cell_region_assignment_99percentile_no_erosion_20190313/cells_heatmap.tif"
for xx in control_dorsal_up
]
obv_du_heatmaps = [
xx +
"/cell_region_assignment_99percentile_no_erosion_20190313/cells_heatmap.tif"
for xx in observers_dorsal_up
]
dmn_du_heatmaps = [
xx +
"/cell_region_assignment_99percentile_no_erosion_20190313/cells_heatmap.tif"
for xx in demonstrator_dorsal_up
]
c = stat.readDataGroup(ctrl_du_heatmaps)
o = stat.readDataGroup(obv_du_heatmaps)
d = stat.readDataGroup(dmn_du_heatmaps)
#comparisons
conds = [
"observer_v_control", "demonstrator_v_observer",
"demonstrator_v_control"
]
comparisons = [[o.astype("float"), c.astype("float")],
[d.astype("float"), o.astype("float")],
[d.astype("float"), c.astype("float")]]
final_dst = pvaldst + "/dorsal_up"
for i, comp in enumerate(comparisons):
print(conds[i])
#set up heatmaps per condition
mm_du_heatmaps = [
os.path.join(src, os.path.join(xx, "cells_heatmap.tif"))
for xx in os.listdir(src) if "mm" in xx
]
fm_du_heatmaps = [
os.path.join(src, os.path.join(xx, "cells_heatmap.tif"))
for xx in os.listdir(src) if "fm" in xx
]
mf_du_heatmaps = [
os.path.join(src, os.path.join(xx, "cells_heatmap.tif"))
for xx in os.listdir(src) if "mf" in xx
]
#read all the heatmaps belonging to each group
mm = stat.readDataGroup(mm_du_heatmaps)
fm = stat.readDataGroup(fm_du_heatmaps)
mf = stat.readDataGroup(mf_du_heatmaps)
#find mean and standard deviation of heatmap in each group
mm_mean = np.mean(mm, axis=0)
mm_std = np.std(mm, axis=0)
fm_mean = np.mean(fm, axis=0)
fm_std = np.std(fm, axis=0)
mf_mean = np.mean(mf, axis=0)
mf_std = np.std(mf, axis=0)
#write mean and standard dev maps to destination
tifffile.imsave(os.path.join(pvaldst, "mm_mean.tif"),
# Voxel-based statistics:
#########################
#Load the data (heat maps generated previously )
group1 = ['/home/yourname/experiment/sample1/cells_heatmap.tif',
'/home/yourname/experiment/sample2/cells_heatmap.tif',
'/home/yourname/experiment/sample3/cells_heatmap.tif',
'/home/yourname/experiment/sample4/cells_heatmap.tif']
group2 = ['/home/yourname/experiment/sample5/cells_heatmap.tif',
'/home/yourname/experiment/sample6/cells_heatmap.tif',
'/home/yourname/experiment/sample7/cells_heatmap.tif',
'/home/yourname/experiment/sample8/cells_heatmap.tif']
g1 = stat.readDataGroup(group1);
g2 = stat.readDataGroup(group2);
#Generated average and standard deviation maps
##############################################
g1a = numpy.mean(g1,axis = 0);
g1s = numpy.std(g1,axis = 0);
g2a = numpy.mean(g2,axis = 0);
g2s = numpy.std(g2,axis = 0);
io.writeData(os.path.join(baseDirectory, 'group1_mean.raw'), rsp.sagittalToCoronalData(g1a));
io.writeData(os.path.join(baseDirectory, 'group1_std.raw'), rsp.sagittalToCoronalData(g1s));
#Load the data (heat maps generated previously )
group1 = [
'/home/yourname/experiment/sample1/cells_heatmap.tif',
'/home/yourname/experiment/sample2/cells_heatmap.tif',
'/home/yourname/experiment/sample3/cells_heatmap.tif',
'/home/yourname/experiment/sample4/cells_heatmap.tif'
]
group2 = [
'/home/yourname/experiment/sample5/cells_heatmap.tif',
'/home/yourname/experiment/sample6/cells_heatmap.tif',
'/home/yourname/experiment/sample7/cells_heatmap.tif',
'/home/yourname/experiment/sample8/cells_heatmap.tif'
]
g1 = stat.readDataGroup(group1)
g2 = stat.readDataGroup(group2)
#Generated average and standard deviation maps
##############################################
g1a = numpy.mean(g1, axis=0)
g1s = numpy.std(g1, axis=0)
g2a = numpy.mean(g2, axis=0)
g2s = numpy.std(g2, axis=0)
io.writeData(os.path.join(baseDirectory, 'group1_mean.raw'),
rsp.sagittalToCoronalData(g1a))
io.writeData(os.path.join(baseDirectory, 'group1_std.raw'),
rsp.sagittalToCoronalData(g1s))
def generate_p_value_maps(src):
"""
generates p-value maps as per ClearMap/analysis.py
#TODO: generalise function
"""
#Load the data (heat maps generated previously )
#make groups
groupA = [
os.path.join(flds, fld) for fld in os.listdir(flds)
if conditions[os.path.basename(fld)] == "homecage_control"
]
groupA.sort()
groupB = [
os.path.join(flds, fld) for fld in os.listdir(flds)
if conditions[os.path.basename(fld)] == "CNO_control_no_reversal"
]
groupB.sort()
groupC = [
os.path.join(flds, fld) for fld in os.listdir(flds)
if conditions[os.path.basename(fld)] == "CNO_control_reversal"
]
groupC.sort()
groupD = [
os.path.join(flds, fld) for fld in os.listdir(flds)
if conditions[os.path.basename(fld)] == "DREADDs"
]
groupC.sort()
group_a = [xx + "/cells_heatmap_60um_erosion.tif" for xx in groupA]
group_b = [xx + "/cells_heatmap_60um_erosion.tif" for xx in groupB]
group_c = [xx + "/cells_heatmap_60um_erosion.tif" for xx in groupC]
group_d = [xx + "/cells_heatmap_60um_erosion.tif" for xx in groupD]
grp_a = stat.readDataGroup(group_a)
grp_b = stat.readDataGroup(group_b)
grp_c = stat.readDataGroup(group_c)
grp_d = stat.readDataGroup(group_d)
#Generated average and standard deviation maps
##############################################
grp_aa = np.mean(grp_a, axis=0)
grp_as = np.std(grp_a, axis=0)
grp_ba = np.mean(grp_b, axis=0)
grp_bs = np.std(grp_b, axis=0)
grp_ca = np.mean(grp_c, axis=0)
grp_cs = np.std(grp_c, axis=0)
grp_da = np.mean(grp_d, axis=0)
grp_ds = np.std(grp_d, axis=0)
io.writeData(os.path.join(src, "group_a_mean.raw"),
rsp.sagittalToCoronalData(grp_aa))
io.writeData(os.path.join(src, "group_a_std.raw"),
rsp.sagittalToCoronalData(grp_as))
io.writeData(os.path.join(src, "group_b_mean.raw"),
rsp.sagittalToCoronalData(grp_ba))
io.writeData(os.path.join(src, "group_b_std.raw"),
rsp.sagittalToCoronalData(grp_bs))
io.writeData(os.path.join(src, "group_c_mean.raw"),
rsp.sagittalToCoronalData(grp_ca))
io.writeData(os.path.join(src, "group_c_std.raw"),
rsp.sagittalToCoronalData(grp_cs))
io.writeData(os.path.join(src, "group_d_mean.raw"),
rsp.sagittalToCoronalData(grp_da))
io.writeData(os.path.join(src, "group_d_std.raw"),
rsp.sagittalToCoronalData(grp_ds))
#Generate the p-values map
##########################
#first comparison
#pcutoff: only display pixels below this level of significance
pvals, psign = stat.tTestVoxelization(grp_a.astype("float"),
grp_d.astype("float"),
signed=True,
pcutoff=0.05)
#color the p-values according to their sign (defined by the sign of the difference of the means between the 2 groups)
pvalsc = stat.colorPValues(pvals, psign, positive=[0, 1], negative=[1, 0])
io.writeData(os.path.join(src, "pvalues_homecage_control_vs_DREADDs.tif"),
rsp.sagittalToCoronalData(pvalsc.astype("float32")))
#second comparison
pvals, psign = stat.tTestVoxelization(grp_a.astype("float"),
grp_b.astype("float"),
signed=True,
pcutoff=0.05)
pvalsc = stat.colorPValues(pvals, psign, positive=[0, 1], negative=[1, 0])
io.writeData(
os.path.join(
src, "pvalues_homecage_control_vs_CNO_control_no_reversal.tif"),
rsp.sagittalToCoronalData(pvalsc.astype("float32")))
#third comparison
pvals, psign = stat.tTestVoxelization(grp_b.astype("float"),
grp_c.astype("float"),
signed=True,
pcutoff=0.05)
pvalsc = stat.colorPValues(pvals, psign, positive=[0, 1], negative=[1, 0])
io.writeData(
os.path.join(
src,
"pvalues_CNO_control_no_reversal_vs_CNO_control_reversal.tif"),
rsp.sagittalToCoronalData(pvalsc.astype("float32")))
#fourth comparison
pvals, psign = stat.tTestVoxelization(grp_c.astype("float"),
grp_d.astype("float"),
signed=True,
pcutoff=0.05)
pvalsc = stat.colorPValues(pvals, psign, positive=[0, 1], negative=[1, 0])
io.writeData(
os.path.join(src, "pvalues_CNO_control_reversal_vs_DREADDs.tif"),
rsp.sagittalToCoronalData(pvalsc.astype("float32")))
os.path.join(pth, os.path.join(xx, subdir + "/cells_heatmap.tif"))
for xx in os.listdir(pth) if "647" in xx and "mouse" in xx
]
ctrl_du_heatmaps.sort()
obv_du_heatmaps = [
os.path.join(pth, os.path.join(xx, subdir + "/cells_heatmap.tif"))
for xx in os.listdir(pth) if "647" in xx and "observ" in xx
]
obv_du_heatmaps.sort()
dmn_du_heatmaps = [
os.path.join(pth, os.path.join(xx, subdir + "/cells_heatmap.tif"))
for xx in os.listdir(pth) if "647" in xx and "demons" in xx
]
dmn_du_heatmaps.sort()
c = stat.readDataGroup(ctrl_du_heatmaps)
o = stat.readDataGroup(obv_du_heatmaps)
d = stat.readDataGroup(dmn_du_heatmaps)
ca = np.mean(c, axis=0)
cstd = np.std(c, axis=0)
oa = np.mean(o, axis=0)
ostd = np.std(o, axis=0)
da = np.mean(d, axis=0)
dstd = np.std(d, axis=0)
#write
io.writeData(os.path.join(pvaldst, "dorsal_up/control_mean.raw"),
rsp.sagittalToCoronalData(ca))