Fixed support for newer Gencode GTF versions #8351
Conversation
|
@jamesemery How "future-proof" is this PR? That is, how likely is it that future releases of Gencode will break the parser again? Has the parser been relaxed to the point where it will tolerate the addition of new fields, etc.? |
|
@droazen I tried to relax absolutely everything but i stopped short of that. I have a test that any of the optional fields we previously special cased can now have any arbitrary value in them (which was the problem that sunk us here). However if they make DRASTIC changes to future gencode releases (like adding new top level transcript types or inventing a new reference orientation than + or -) then all bets are off... |
...roadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java
Show resolved
Hide resolved
...a/org/broadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotation.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/utils/codecs/gtf/GencodeGtfCodec.java
Outdated
Show resolved
Hide resolved
...roadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/utils/codecs/gtf/GencodeGtfFeature.java
Show resolved
Hide resolved
src/main/java/org/broadinstitute/hellbender/utils/codecs/gtf/GencodeGtfFeature.java
Outdated
Show resolved
Hide resolved
…from GencodeGTFFeature (which had a bug with ordering before)
jamesemery
force-pushed
the
je_fixGTFCodecForGencodeV43Support
branch
from
58d2262
to
125aa5a
Compare
|
@jonn-smith responded to comments and back to you. I fully got rid of the old unparsed string of "anonymousOptionalFields" that were occasionally relevant for non-blessed fields in gencode. There is an enum list of the known optional fields for gencode but everything gets parsed into the same key-value list which should make phase 2 easier to finish here. |
Fixes #7166 #7385
This doesn't contain any of the necessary work to support Gencode GFF3 files yet #, that will (probably) come in a subsequent PR as it requires a much more substantial refactoring effort of the Gencode datasources code.