Genome Nexus changes the Reference Allele and Tumor_Seq_Allele2

[thomasyu888]

• Input
  input.txt

• annotation-tools intermediate files I must add the .txt at the end or github won't allow me to upload these. My understanding it the input.txt.temp.annotated.txt is the output from Genome Nexus. But because the annotation-tools allows us to include a directory with a list of mafs or vcfs, it annotates each one of those files separately. processed.txt is all of these merged.
  input.txt.temp.annotated.txt
  input.txt.temp.txt

• Processed
  processed.txt

[inodb]

Maybe not related but looks likeI end_position column is missing in the input? Not sure if that's the reason things go haywire tho

[thomasyu888]

Interesting. The input file doesn't have End_Position because we don't require centers to include it, but I made an edit to the original post to include End_Position.

[sheridancbio]

This appears (on first glance) to be related to the issue of a shared common prefix in the alleles. A replacement of "A" with "AT" is (I think) semantically equivalent to an insertion of a "T" after the referenced "A". But the output above seems to be a bug of some sort. If a removal of a shared prefix had occurred, the offsets should have been adjusted. The first line has a shared prefix of CTTTTTTTTTTTT followed by either a T [reference] or a C [tumor]. Isn't this semantically equivalent to a SNP at the final position? So in that case , it should be a T->C SNP at position (170837513+13). I also notice that there are three input records here, but only 2 output records (post processing) I wonder what happened there. @ao508 may know these rules off the top of her head.

[thomasyu888]

Thanks @sheridancbio and @inodb . I made edits to my initial post - that is the complete list.

[ao508]

@thomasyu888 Would you mind slacking me the input MAFs with all of the fields? That would make it easier for me to troubleshoot the issue with the MAF standardization script we provided.

[thomasyu888]

@inodb @ao508 , This is ready to look at

I took a quick glance at the annotation-tools file before it gets sent to Genome Nexus, it appears that this could be a Genome Nexus issue.

[inodb]

Fixed in #39