def process_image(tiff_dir, mask_file, d):
# read and aggregate data
data = aggregate_image(tiff_dir, mask_file)
X, exp = data['X'], data['exp']
exp = filter_out(exp)
# Get total counts per cell
tot = pd.DataFrame(exp.sum(1))
tot.columns = ['total_count']
# remove cells with total count bellow 3
X = X[tot.values > 3]
exp = exp[tot.values > 3]
tot = tot[tot.values > 3]
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(exp.T).T
res = NaiveDE.regress_out(tot, dfm.T, 'np.log(total_count)').T
# Add total_count as pseudogene for reference
# res['log_total_count'] = np.log(tot['total_count'])
res.to_csv(d+'/expressions.txt', sep=' ', header=True, index=False)
X.to_csv(d+'/positions.txt', sep=',', header=False, index=False)
def main():
df = pd.read_table('data/Layer2_BC_count_matrix-1.tsv', index_col=0)
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
sample_info = get_coords(df.index)
sample_info['total_counts'] = df.sum(1)
sample_info = sample_info.query(
'total_counts > 5') # Remove empty features
df = df.loc[sample_info.index]
X = sample_info[['x', 'y']]
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
# Add total_count as pseudogene for reference
res['log_total_count'] = np.log(sample_info['total_counts'])
results = SpatialDE.run(X, res)
sample_info.to_csv('BC_sample_info.csv')
results.to_csv('BC_final_results.csv')
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv('BC_MS_results.csv')
return results
def process_mer_fish(exp_file, annotation_file, d):
# read data
df = pd.read_csv(exp_file, index_col=0)
annotations = pd.read_csv(annotation_file)['gene']
# aggregate data
tmp = aggregate_data(df, annotations)
X, exp = tmp['X'], tmp['exp']
# filter practically unobserved genes
exp = exp.T[exp.sum(0) >= 3].T
# Get total counts per cell
tot = pd.DataFrame(exp.sum(1))
tot.columns = ['total_count']
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(exp.T).T
res = NaiveDE.regress_out(tot, dfm.T, 'np.log(total_count)').T
# Add total_count as pseudogene for reference
res['log_total_count'] = np.log(tot['total_count'])
res.to_csv(d+'/expressions.txt', sep=' ', header=True, index=False)
X.to_csv(d+'/positions.txt', sep=',', header=False, index=False)
def main():
# Get time points for each sample
sample_info = pd.read_csv('Frog_sample_info.csv', index_col=0)
# Load expression
df = pd.read_csv('data/GSE65785_clutchApolyA_relative_TPM.csv',
index_col=0)
df = df[sample_info.index]
df = df[df.sum(1) >= 3] # Filter practically unobserved genes
X = sample_info[['hpf']]
# Convert expression data to log scale, with genes in columns
dfm = NaiveDE.stabilize(df)
res = NaiveDE.regress_out(sample_info,
dfm,
'np.log(ERCC) + np.log(num_genes)',
rcond=1e-4).T
# Add technical factors as pseudogenes for reference
res['log_num_genes'] = np.log(sample_info['num_genes'])
res['log_ERCC'] = np.log(sample_info['ERCC'])
# Perform Spatial DE test with default settings
results = SpatialDE.run(X, res)
# Save results and annotation in files for interactive plotting and interpretation
results.to_csv('Frog_final_results.csv')
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv('Frog_MS_results.csv')
return results
def main():
df = pd.read_csv('exp_mat_43.csv', index_col=0)
df.columns = df.columns.map(int)
# Get coordinates for each sample
sample_info = pd.read_csv('sample_info_43.csv', index_col=0)
df = df[sample_info.index]
X = sample_info[['x', 'y']]
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm, 'np.log(total_count)').T
# Add total_count as pseudogene for reference
res['log_total_count'] = np.log(sample_info['total_count'])
# Perform Spatial DE test with default settings
results = SpatialDE.run(X, res)
# Save results and annotation in files for interactive plotting and interpretation
results.to_csv('final_results_43.csv')
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv('MS_results_43.csv')
return results
def main():
df = pd.read_csv('10t.csv', index_col=0)
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
# Get coordinates for each sample
sample_info = get_coords(df.index)
sample_info['total_counts'] = df.sum(1)
sample_info = sample_info.query(
'total_counts > 10') # Remove empty features
df = df.loc[sample_info.index]
X = sample_info[['x', 'y']]
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
# Add total_count as pseudogene for reference
res['log_total_count'] = np.log(sample_info['total_counts'])
# Perform Spatial DE test with default settings
results = SpatialDE.run(X, res)
# Save results and annotation in files for interactive plotting and interpretation
sample_info.to_csv('10t_sample_info.csv')
results.to_csv('10t_final_results.csv')
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv('10t_MS_results.csv')
return results
def main():
df = pd.read_csv('data/rep6/middle_exp_mat.csv', index_col=0)
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
# Get coordinates for each sample
sample_info = pd.read_csv('data/rep6/middle_sample_info.csv', index_col=0)
df = df.loc[sample_info.index]
X = sample_info[['abs_X', 'abs_Y']]
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_count)').T
# Add total_count as pseudogene for reference
res['log_total_count'] = np.log(sample_info['total_count'])
# Perform Spatial DE test with default settings
results = SpatialDE.run(X, res)
# Assign pi_0 = 1 in multiple testing correction
results['qval'] = SpatialDE.util.qvalue(results['pval'], pi0=1.0)
# Save results and annotation in files for interactive plotting and interpretation
sample_info.to_csv('middle_sample_info.csv')
results.to_csv('middle_final_results.csv')
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv('middle_MS_results.csv')
return results
def main(expression_csv, coordinate_csv, results_csv, model_selection_csv):
''' Perform SpatialDE test on data in input files.
<expression csv> : A CSV file with expression valies. Columns are genes,
and Rows are samples
<coordinates csv> : A CSV file with sample coordinates. Each row is a sample,
the columns with coordinates must be named 'x' and 'y'. For other formats
(e.g. 1d or 3d queries), it is recommended to write a custom Python
script to do the analysis.
<output file> : P-vaues and other relevant values for each gene
will be stored in this file, in CSV format.
'''
df = pd.read_csv(expression_csv, index_col=0)
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
sample_info = pd.read_csv(coordinate_csv, index_col=0)
sample_info['total_counts'] = df.sum(1)
sample_info = sample_info.query('total_counts > 5') # Remove empty features
df = df.loc[sample_info.index]
X = sample_info[['x', 'y']]
# Convert data to log-scale, and account for depth
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
# Perform Spatial DE test with default settings
results = SpatialDE.run(X, res)
# Save results and annotation in files for interactive plotting and interpretation
results.to_csv(results_csv)
if not model_selection_csv:
return results
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, res, de_results)
ms_results.to_csv(model_selection_csv)
return results, ms_results
def spatialde_test(adata,
coord_columns=['x', 'y'],
regress_formula='np.log(total_counts)'):
''' Run the SpatialDE test on an AnnData object
Parameters
----------
adata: An AnnData object with counts in the .X field.
coord_columns: A list with the columns of adata.obs which represent spatial
coordinates. Default ['x', 'y'].
regress_formula: A patsy formula for linearly regressing out fixed effects
from columns in adata.obs before fitting the SpatialDE models.
Default is 'np.log(total_counts)'.
Returns
-------
results: A table of spatial statistics for each gene.
'''
logging.info('Performing VST for NB counts')
adata.layers['stabilized'] = NaiveDE.stabilize(adata.X.T).T
logging.info('Regressing out fixed effects')
adata.layers['residual'] = NaiveDE.regress_out(
adata.obs, adata.layers['stabilized'].T, regress_formula).T
X = adata.obs[coord_columns].values
expr_mat = pd.DataFrame.from_records(adata.layers['residual'],
columns=adata.var.index,
index=adata.obs.index)
results = run(X, expr_mat)
# Clip 0 pvalues
min_pval = results.query('pval > 0')['pval'].min() / 2
results['pval'] = results['pval'].clip_lower(min_pval)
# Correct for multiple testing
results['qval'] = qvalue(results['pval'], pi0=1.)
return results
def Spatial_DE_AEH(filterd_exprs,coordinates,results,pattern_num,l = 1.05, verbosity = 1):
## Automatic expression histology
coordinates_cp =coordinates.copy()
coordinates_cp['total_counts'] = filterd_exprs.sum(1)
dfm = NaiveDE.stabilize(filterd_exprs.T).T
res = NaiveDE.regress_out(coordinates_cp, dfm.T, 'np.log(total_counts)').T
results['pval'] = results['pval'].clip(lower = results.query('pval > 0')['pval'].min() / 2)
results['qval'] = results['qval'].clip(lower = results.query('qval > 0')['qval'].min() / 2)
sres = results.query('qval < 0.05 & g != "log_total_count"').copy()
X = coordinates.values
histology_results, patterns = SpatialDE.spatial_patterns(X, res, sres, int(pattern_num), l = l,verbosity=verbosity)
pattern_dic = {"histology_results":histology_results,"patterns":patterns}
return pattern_dic
def main():
sample_info = pd.read_csv('MOB_sample_info.csv', index_col=0)
df = pd.read_csv('data/Rep11_MOB_0.csv', index_col=0)
df = df.loc[sample_info.index]
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
X = sample_info[['x', 'y']].values
times = pd.DataFrame(columns=['N', 'time'])
Ns = [50, 100, 200, 300, 500, 750, 1000, 2000]
j = 0
for N in Ns:
for i in range(5):
Y = res.sample(N, axis=1).values.T
t0 = time()
m = GPclust.MOHGP(X=X,
Y=Y,
kernF=kern.RBF(2) + kern.Bias(2),
kernY=kern.RBF(1) + kern.White(1),
K=5,
prior_Z='DP')
m.hyperparam_opt_args['messages'] = False
m.optimize(step_length=0.1, verbose=False, maxiter=2000)
times.loc[j] = [N, time() - t0]
print(times.loc[j])
j += 1
times.to_csv('AEH_times.csv')
def main(out_file):
df = pd.read_table('../../BreastCancer/data/Layer2_BC_count_matrix-1.tsv',
index_col=0)
df = df.T[df.sum(0) >= 3].T # Filter practically unobserved genes
sample_info = get_coords(df.index)
sample_info['total_counts'] = df.sum(1)
sample_info = sample_info.query(
'total_counts > 5') # Remove empty features
# Bootstrap sampling 80% of data
sample_info = sample_info.sample(frac=0.8)
df = df.loc[sample_info.index]
X = sample_info[['x', 'y']]
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
results = SpatialDE.run(X, res)
results.to_csv(out_file)
return results
def Spatial_DE(filterd_exprs, coordinates):
if(filterd_exprs.shape[0] != coordinates.shape[0]):
sys.exit("The number of cells in expression file and location file don't match\n")
else:
## results and ms_results
coordinates_cp = coordinates.copy()
coordinates_cp['total_counts'] = filterd_exprs.sum(1)
dfm = NaiveDE.stabilize(filterd_exprs.T).T
res = NaiveDE.regress_out(coordinates_cp, dfm.T, 'np.log(total_counts)').T
res['log_total_count'] = np.log(coordinates_cp['total_counts'])
results = SpatialDE.run(coordinates, res)
de_results = results[(results.qval < 0.05)].copy()
if(de_results.shape[0] > 0):
ms_results = SpatialDE.model_search(coordinates, res, de_results)
result_dic = {"results":results, "ms_results":ms_results}
else:
print("No spatially variable genes found! \n")
result_dic = {"results":results}
return result_dic
import glob
samp_counts = glob.glob(
'/fastscratch/myscratch/shicks1/HumanPilot/sample_data/by_sample_id/*_counts.csv'
)
samp_meta = np.ravel([[x[:-10] + 'meta.csv'] for x in samp_counts])
samp_output = np.ravel([[x[:-10] + 'spatialDE_results.csv']
for x in samp_counts])
df = pd.DataFrame({
'counts': samp_counts,
'meta': samp_meta,
'out': samp_output
})
for index, row in df.iterrows():
counts = pd.read_csv(row['counts'], index_col=0) # load counts
counts = counts.T[counts.sum(
axis=0) >= 5].T # Filter practically unobserved genes
sample_info = pd.read_csv(
row['meta'], index_col=0) # load meta data with spatial coordinates
norm_expr = NaiveDE.stabilize(counts.T).T # remove tech variation
resid_expr = NaiveDE.regress_out(sample_info, norm_expr.T, 'np.log(sum)').T
X = sample_info[['imagerow', 'imagecol']]
now = datetime.now().strftime("%H:%M:%S")
print(now)
results = SpatialDE.run(X, resid_expr)
now = datetime.now().strftime("%H:%M:%S")
print(now)
results.to_csv(row['out']) # Save spatial results
print("Finished =", row['counts'])
import pandas as pd
import NaiveDE
import SpatialDE
counts = pd.read_csv('./processed_data/MERFISH_Animal18_Bregma0.11_countdata.csv', index_col=0)
counts = counts.T[counts.sum(0) >= 3].T
sample_info = pd.read_csv('./processed_data/MERFISH_Animal18_Bregma0.11_info.csv', index_col=0)
sample_info['total_counts'] = counts.sum(1)
counts = counts.loc[sample_info.index]
norm_expr = NaiveDE.stabilize(counts.T).T
resid_expr = NaiveDE.regress_out(sample_info, norm_expr.T, 'np.log(total_counts)').T
sample_resid_expr=resid_expr
X = sample_info[['x', 'y']]
results = SpatialDE.run(X, sample_resid_expr)
results.to_csv('./output/MERFISH_Animal18_Bregma0.11_spe.csv',sep=' ', index=False, header=True)
de_results = results[(results.qval < 0.05)].copy()
ms_results = SpatialDE.model_search(X, resid_expr, de_results)
ms_results.to_csv('./output/MERFISH_Animal18_Bregma0.11_ms_spe.csv',sep=' ', index=False, header=True)
import numpy as np
import NaiveDE
import SpatialDE
import time
info = pd.read_csv("../processed_data/Rep11_MOB_info_scgco.csv", index_col=0)
exp_diff = 1
for noise in [0.1, 0.2, 0.3, 0.4, 0.5, 0.6]:
for irep in range(10):
ff = "../processed_data/sim_MOB_expdiff" + str(
exp_diff) + "_noise" + str(noise) + "_counts" + str(irep) + ".csv"
print(ff)
df = pd.read_csv(ff, index_col=0)
df = df.T[df.sum(0) >= 3].T
sample_info = info.copy()
X = sample_info[['x', 'y']]
start_time = time.time()
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
res['log_total_count'] = np.log(sample_info['total_counts'])
results = SpatialDE.run(X, res)
ff = "../spatialde_results/sim_MOB_expdiff" + str(
exp_diff) + "_noise" + str(noise) + "_counts" + str(
irep) + "_spe.csv"
results.to_csv(ff)
def main():
df = pd.read_csv('10t.csv', index_col=0)
df = df.T[df.sum(0) >= 3].T
sample_info = get_coords(df.index)
sample_info['total_counts'] = df.sum(1)
sample_info = sample_info.query(
'total_counts > 10') # Remove empty features
df = df.loc[sample_info.index]
# X = sample_info[['x', 'y']]
dfm = NaiveDE.stabilize(df.T).T
res = NaiveDE.regress_out(sample_info, dfm.T, 'np.log(total_counts)').T
res['log_total_count'] = np.log(sample_info['total_counts'])
results = pd.read_csv('10t_final_results.csv', index_col=0)
results['pval'] = results['pval'].clip_lower(
results.query('pval > 0')['pval'].min() / 2)
results['qval'] = results['qval'].clip_lower(
results.query('qval > 0')['qval'].min() / 2)
ymy = int(sys.argv[1])
sres = results.query('qval < 0.05 & g != "log_total_count"').copy()
#a = sres['l'].value_counts()
#a.to_csv('10t_l_results.csv')
X = sample_info[['x', 'y']].values
histology_results, patterns = SpatialDE.spatial_patterns(X,
res,
sres,
ymy,
11,
verbosity=1)
histology_results.to_csv('10t_AEH_results.{}.csv'.format(ymy))
patterns.add_prefix('pattern_').to_csv(
'10t_pattern_results.{}.csv'.format(ymy))
for i, Ci in enumerate(
histology_results.sort_values('pattern').pattern.unique()):
fig = plt.figure(figsize=(5, 5))
plt.scatter(sample_info['x'],
sample_info['y'],
c=patterns[Ci],
s=10,
cmap=plt.get_cmap("YlOrBr"),
edgecolor="none",
marker='s')
plt.axis([0, 50, 0, 50])
plt.xlim(0, 50)
plt.ylim(0, 50)
plt.xticks([0, 10, 20, 30, 40, 50])
plt.yticks([0, 10, 20, 30, 40, 50])
plt.axis('equal')
plt.gca().invert_yaxis()
plt.title('Pattern {} - {} genes'.format(
i,
histology_results.query('pattern == @i').shape[0]),
size=20)
plt.tight_layout()
plt.savefig("10t.{}.{}.pdf".format(ymy, i), bbox_inches='tight')
for i in histology_results.sort_values('pattern').pattern.unique():
print('Pattern {}'.format(i))
print(', '.join(
histology_results.query('pattern == @i').sort_values('membership')
['g'].tolist()))
print()
return histology_results
import pandas as pd import numpy as np import NaiveDE,SpatialDE from somde import SomNode dataname = '../slideseq_data/Puck_180819_11_' df = pd.read_csv(dataname+'count.csv',sep=',',index_col=1) corinfo = pd.read_csv(dataname+'idx.csv',sep=',',index_col=0) del(df['ENSEMBL']) print(df.shape) corinfo["total_count"]=df.sum(0) # stablize,regress_out is gene by cell . However, run is cell by gene dfm = NaiveDE.stabilize(df) res = NaiveDE.regress_out(corinfo, dfm, 'np.log(total_count)').T X=corinfo[['x','y']].values.astype(np.float32) som4 = SomNode(X,20) ndf,ninfo = som4.mtx(df) r1 ,numberq =som4.run() nres = som4.norm() som4.view()
