As part of the indel workflow we should implement a nonsense mediated decay (NMD) checking step. This paper gives a good description of the process + why it is important to include a NMD check for indel and alternative splicing originating neoantigens.
I have been looking into a few options.
A plugin for VEP. This option could be very easy to implement but may not work with TEMPO's current VEP version (88). We maybe would need to change this version or ask the genome nexus folk to set up the plugin as we may be annotating variants through their pipeline in the future.
aenmd, an R package that annotates for NMD that says it is better than the VEP plugin. Takes VCF as an input, which we would likely need to change for taking a MAF. They also have a CLI which is nice.
We make one ourselves. Doing a simple one looks relatively easy and could be done in python. The VEP annotation is really just 3 rules. aenmd may implements 5 rules and they're all listed in the paper. If we go this rule based route then perhaps implementing it ourselves could be an option
There are other tools out there that are more complex. I'm still looking into those.
Description of feature
As part of the indel workflow we should implement a nonsense mediated decay (NMD) checking step. This paper gives a good description of the process + why it is important to include a NMD check for indel and alternative splicing originating neoantigens.
I have been looking into a few options.
A plugin for VEP. This option could be very easy to implement but may not work with TEMPO's current VEP version (88). We maybe would need to change this version or ask the genome nexus folk to set up the plugin as we may be annotating variants through their pipeline in the future.
aenmd, an R package that annotates for NMD that says it is better than the VEP plugin. Takes VCF as an input, which we would likely need to change for taking a MAF. They also have a CLI which is nice.
We make one ourselves. Doing a simple one looks relatively easy and could be done in python. The VEP annotation is really just 3 rules. aenmd may implements 5 rules and they're all listed in the paper. If we go this rule based route then perhaps implementing it ourselves could be an option
There are other tools out there that are more complex. I'm still looking into those.