`nucpos` files are empty

[sameet]

Hi, I did the following:

1. Ran bwa mem for the alignment.
2. Used macs2 to call peaks (broad peaks).
3. Used the NucleoATAC pipeline as given in the help documents.

The pipeline ran without errors. However, the nucpos output was empty. It still generated other files that are non-empty, including nfr files that also seem to have data. Am I missing something?

[AliciaSchep]

Please see Issue #55 and discussion there -- might be same problem.

[sameet]

Hi @AliciaSchep, I looked at Issue #55, and tried the following. Briefly, I deleted the .pyxbld from my home directory. Then i used python tests.py. It forced re-compile of the C part of the code. I used the GCC 5.4.0. I still get the following error:

======================================================================
FAIL: Make sure variance calculation is close to what is obtained by simulation
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/gpfs/ycga/home/sm2556/GitHub/NucleoATAC/tests/test_var.py", line 33, in test_sd1 
    self.assertTrue(abs(sd1-sd2)<0.05*sd1)
AssertionError: False is not true

======================================================================
FAIL: Make sure variance calculation is same as would be obtained through alternate calculation
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/gpfs/ycga/home/sm2556/GitHub/NucleoATAC/tests/test_var.py", line 43, in test_sd2
    self.assertTrue(abs(sd1-sd2)<0.001*sd1)
AssertionError: False is not true

----------------------------------------------------------------------
Ran 24 tests in 205.134s

My python is 2.7.13. What should be my next step?

[sm2556@c13n06 NucleoATAC]$ strings -a ~/.pyxbld/lib.linux-x86_64 2.7/nucleoatac/multinomial_cov.so | grep "GCC: ("
GCC: (GNU) 5.4.0

The problem persists.