Filter variants by gene symbol

[KoalaQin]

This changed back to using Browser tables, since they had multi-step processing of the Gencode table. A few reasons to use their gene table instead of our gencode table:

1. they retrieved gene symbol from HGNC and get the symbols from Gencode gene name if there's no symbol there (however, there are ~1924 symbols aren't equal between HGNC and Gencode, should we support gene ID instead?)
2. they merged the intervals for CDS, UTR and exons step by step;
3. already removed PAR_Y genes;

We had discussion on the display of exon-centric variants in coding and non-coding genes.
Note: this code is a bit slow, for DRD2, it takes ~40s to finish locally.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[jkgoodrich]

Can you determine a way to make this run faster? i think we absolutely need this to be able to run locally in a time as fast as the time used in the current code. Is there a place these differences are clearly outlined? I don't like the idea of having resources that differ like this without clear documentation of what the differences are, and ideally a HT that can map exactly from one to the other. For instance, in our research we use this gencode HT, is this indicating that we should not be?

[KoalaQin]

The gene symbol shouldn't be a big issue, they may have reason to prefer HGNC symbols and they do put Symbol in GENCODE v35 (though it's a typo, should be v39) on the gene page, like here.
Maybe we should check, without merging overlapping intervals, even for non-protein-coding genes, if we get the same number of variants.