error with NA is size factors using ACCOST

[tuxette]

Due to the error reported in #2, I tried to run ACCOST on plain text file (uncompressing .gz files in example):

python ACCOST.py 40000 example/bins_chr.txt example/info.tab hIMR90 hESC res3

and had the following error:

Welcome. Starting ACCOST.
Output will be written to res3
temp directory currently set to: /tmp/tmplcecgahe
Reading bin midpoints and mappability from example/bins_chr.txt
Found 3347 bins, including 113 low mappability bins.
Read 1 pairs of files from example/info.tab.
Loading data
Loading biases from example/hIMR90_chr12_r1.biases
Loading biases from example/hESC_chr12_r1.biases
Found 2 matrix files.
Total bins: 3347
Sorting data by genomic distance
Total counts in example/hIMR90_chr12_r1 is 3286121.
Total counts in example/hESC_chr12_r1 is 832449.
Calculating size factors
calculating size factors: nBins=3347 nDistances=3347 nReplicates=2
/work/nvilla/chrocodiff/accost/contact_counts.py:521: RuntimeWarning:
Mean of empty slice
   mean = np.nanmean(c,axis=1)
/tools/python/3.6.3/lib/python3.6/site-packages/numpy/lib/nanfunctions.py:1114:
RuntimeWarning: All-NaN slice encountered
   overwrite_input=overwrite_input)
Filling nans in size factors with 1
Calculating means and variances
/work/nvilla/chrocodiff/accost/running_stats.py:10: RuntimeWarning: Mean
of empty slice
   result[i] = np.nanmean(values[i:i + window,:])
/work/nvilla/chrocodiff/accost/running_stats.py:18: RuntimeWarning:
Degrees of freedom <= 0 for slice.
   result[i] = np.nanvar(values[i:i + window,:])
Traceback (most recent call last):
   File "ACCOST_internal.py", line 463, in <module>
     main()
   File "ACCOST_internal.py", line 416, in main
     matrixA =
contact_counts.combine_replicates_and_calculate_mean_variance(mats_A,smooth_dist)
   File "/work/nvilla/chrocodiff/accost/contact_counts.py", line 751, in
combine_replicates_and_calculate_mean_variance
     means,variances,z_factors,corrected_variances =
calc_mean_var(readers,all_biases,all_size_factors,nBins,nDistances,badBins,smooth_dist)
   File "/work/nvilla/chrocodiff/accost/contact_counts.py", line 619, in
calc_mean_var
     if i_values[current_distance].shape[0]<= ii or
np.isnan(i_values[current_distance][ii][0]):
IndexError: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis
(`None`) and integer or boolean arrays are valid indices
Traceback (most recent call last):
   File "annotate_pvals.py", line 51, in <module>
     main()
   File "annotate_pvals.py", line 22, in main
     pvalfh = open(pval_file,'r')
FileNotFoundError: [Errno 2] No such file or directory: 'res3/_ln_pvals.txt'
ACCOST Done!

Deactivating the computation of size factors with --no_dist_norm leads to a similar error in variance computation.