Add read_evidence module

[timodonnell]

• Add read_evidence module, tests, and test data. This is quite slow, and there are several optimization opportunities, but I wanted to get the API settled first.
• Add locus module.
• Add pysam dependency.
• Add all_standard_nucleotides function in nucleotides module.

[iskandr]

Comments from the review on Reviewable.io

---

^{varcode/locus.py, line 30 [r1] (raw file):}
Does the Python 3 vs. Python 2 difference in generator vs. list matter in how this gets used?

---

^{varcode/locus.py, line 39 [r1] (raw file):}
Isn't str(self) redundant with the "%s" format string?

---

^{varcode/read_evidence.py, line 64 [r1] (raw file):}
Could you add an example of what's actually in the tuple?

---

^{varcode/read_evidence.py, line 286 [r1] (raw file):}
What do you think of moving this to its own file? And maybe also making pileup.py?

---

^{varcode/read_evidence.py, line 574 [r1] (raw file):}
Why not use a collections.Counter?

---

[timodonnell]

Comments from the review on Reviewable.io

---

^{varcode/locus.py, line 30 [r1] (raw file):}
Yeah, I guess it depends on how it gets used. My intention is to not rely on it being a list. I'm not sure how to enforce that though.

---

^{varcode/locus.py, line 39 [r1] (raw file):}
Done.

---

^{varcode/read_evidence.py, line 64 [r1] (raw file):}
I have a feeling that bit may change a bunch and would rather not duplicate that info around the codebase. I'll add a pointer in the doc to the function that generates the tuple though, so someone could find out themselves easily.

---

^{varcode/read_evidence.py, line 286 [r1] (raw file):}
Sure, refactored into submodule.

---

^{varcode/read_evidence.py, line 574 [r1] (raw file):}
Made this clearer with a comment and by pulling out the lambda into a def.

---

[iskandr]

Comments from the review on Reviewable.io

---

LGTM

---