Feat speed optimization

[taoliu]

1. Remove hashtable.pyx and use py3 dict for pvalue-qvalue checkup Feat: Optimize MACS speed in Python3 #335

   After testing the CPU/mem usage in v2.1.4 (py2.7) and v2.2.4 (py3.7), I found that the old hashtable.pyx implementation copied from Pandas (very old version) doesn't work well in Python3. It slows down the pvalue-qvalue checkup (in CallPeakUnit.pyx) for about 40 times with the identical Cython codes. While testing on my laptop on 5million ChIP vs 5 million Control (newly added testing data in test/ folder), the getitem function in the hashtable.pyx took 3.5s with 37,382,037 calls in MACS2 v2.1.4, but 148.6s with the same number of calls in MACS2 v2.2.4. Therefore, I fall back to the standard python dictionary implementation for pqtable checkup. It is faster than the old py2 version hashtable.pyx, but uses a bit more memory. As an example, with the new implementation (py3 dict) in the branch feat_speed_optimization can finish 5M reads test 20% faster than MACS2 v2.1.4, and use 15% more memory than MACS2 v2.1.4.

2. Add 5Million reads ChIP (CTCF from ENCODE2) and Ctrl for testing MACS2 performance. Now the test.sh will output a summary of CPU and memory usage of the 5M test run.

PS: May relate to #334

[codecov]

Codecov Report

Merging #338 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #338   +/-   ##
=======================================
  Coverage   94.54%   94.54%           
=======================================
  Files           8        8           
  Lines         440      440           
=======================================
  Hits          416      416           
  Misses         24       24

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0e871a8...46ec485. Read the comment docs.