Would you want VEP or snpEff or something else? Is it okay if PheWeb only gives users one choice?
Annotation: I want to be quite careful not to introduce problems into sites.tsv, since matrixify has to parse it. Check version numbers of VEP/snpEff, and then have PheWeb manage everything. If VEP or snpEff can add exactly the annotations I want to a vcf-like file, this should be easy– I just annotate sites.tsv, pass along all columns into augmented_phenos_gz/*.gz and matrix.tsv.gz, convince my API to load and serve the annotations, and set up my default templates to render whatever's available.
From this, I would display consequence: stopgained in tables and maybe use IMPACT: HIGH to color Manhattan/LZ. Most of the rest could get dumped at the bottom of the PheWAS page. If I'm happy with the way SYMBOL and Feature are decided, I could use those instead of my gene annotation.
As I read the VEP annotation from a VCF, I'll take the most deleterious consequence and the highest IMPACT. Then assert that CSQ["alt"] == alt.
Later: Figure out whether we can run ensembl-vep on sites.tsv, or if not what I must add. Maybe add an empty INFO column or rename rsids to ID.
(extracted from another issue)
Sarah: add an annotation column (for example: intronic; exonic-nonsynonymous; exonic-splice; exonic-synonynmous; intergenic; 5' UTR; 3' UTR)
Peter:
Annotation: I want to be quite careful not to introduce problems into
sites.tsv
, sincematrixify
has to parse it. Check version numbers of VEP/snpEff, and then have PheWeb manage everything. If VEP or snpEff can add exactly the annotations I want to a vcf-like file, this should be easy– I just annotatesites.tsv
, pass along all columns intoaugmented_phenos_gz/*.gz
andmatrix.tsv.gz
, convince my API to load and serve the annotations, and set up my default templates to render whatever's available.Sarah: VEP only is fine.
Annotation: VEP can write into a VCF file (--vcf flag) http://useast.ensembl.org/info/docs/tools/vep/script/vep_options.html#opt_vcf You can can filter the VCF output http://useast.ensembl.org/info/docs/tools/vep/script/vep_filter.html
Peter: VEP adds this to the end of
INFO
:documented by this header:
in pairs:
From this, I would display
consequence: stopgained
in tables and maybe useIMPACT: HIGH
to color Manhattan/LZ. Most of the rest could get dumped at the bottom of the PheWAS page. If I'm happy with the waySYMBOL
andFeature
are decided, I could use those instead of my gene annotation.As I read the VEP annotation from a VCF, I'll take the most deleterious
consequence
and the highestIMPACT
. Then assert thatCSQ["alt"] == alt
.Later: Figure out whether we can run
ensembl-vep
onsites.tsv
, or if not what I must add. Maybe add an emptyINFO
column or renamersids
toID
.Sarah:
add to tables:
add to tooltips: