Mutect2 error: Badly formed genome unclippedLoc

Question

[aldhairmedico]

Hi everyone,

I'm trying to run Mutect2 for WES cancer data.

However, since their Resource bundle only supports h19 seems I cannot proceed (I want to compare it with Strelka2 results).

I've been looking for some hg38 interval_list file and I found: hg38_v0_HybSelOligos_whole_exome_illumina_coding_v1_whole_exome_illumina_coding_v1.Homo_sapiens_assembly38.targets.interval_list

However, when I run the GenomicsDBImport I get the error (no matter if I use my own hg38 reference and .dict or the ones from GATK Resource Bundle):

A USER ERROR has occurred: Badly formed genome unclippedLoc: Contig chr1 given as location, but this contig isn't present in the Fasta sequence dictionary

So, my questions are:

1. Does anyone know if there is any release date for this hg38 based exome interval file?
2. Or the file I put is ok and the error is coming from somewhere else?

[z.jalali.96]

Hi

I have the same problem in mutect2. I changed my reference file several times, using NCBI, ensemble and broadinstitute google bundle but still face the same error. I made the dictionary using either picard and gatk CreateSequenceDictionary. but still get the error. Do you know what might be the problem? I would appreciate any input on this, feel frustrated.

Thank you so much

[aldhairmedico]

I strongly recommend you download the FAI and DICT files from the GATK Resource bundle as well. I had the same errors with the dictionary that I made by myself.

[376357153]

may be "chr1" in your reference call "1"

[Ram]

OP has already mentioned what their error was. I'm moving your answer to a comment.

Answer 1

[aldhairmedico]

For the ones interested:

This error was a product of incorrect indexing of Fasta reference.

I used the hg38 reference available on the NCBI web.

However, all the chromosome regions are listed as NC_10000.0 codes instead of 1, 2, 3 or chr1 chr2 and chr3.

I had to change it to another reference (I downloaded both .FA and .FAI from GATK Resource Bundle but it seems to work fine with hg38 from Ensembl too).