Covariate file can be empty?

[nunofonseca]

yep.
UserWarning: loadtxt: Empty input file: "fqtl/cov.tsv"

[clody23]

the tsv file with covariates is used by get_covariates.py script.
For the moment the script generates an hdf5 file with two groups:
1 - array of covariates
2 - array of sample names

This hdf5 file is then used by three other scripts (sort_ids.py, runpeer.py and cis/tran_eqtl.py).
If we want to change this, should the logic be like:

get_covariates.py :
if cov_file.txt is empty
then generate a fake covariates.hdf5 file

sort_ids.py:
if covariates.hdf5 file is fake
then output a file with the same name (*sorted.hdf5) but which is still fake
else sort the file.

runpeer.py:
if (for obscure reasons since the file is empty) the user select peer_cov=yes
then check if the covariates.hdf5 file is fake and print something to say that covariates will not be considered in the model
else pass

cis/tran_eqtl.py (here is complicated, but there can be a trick):
if covariates.hdf5 is fake
then peer_cov=yes (this option disable usage of covariate in the lmm model)
else pass

???