Error: Have found no compatible reference specifications

[santhoshnh]

Hii,
I installed HLA-LA through conda. I have downloaded graph and indexed, And while running the HLA-LA.pl I am getting below error

Have found no compatible reference specifications in /home/admin/anaconda3/opt/hla-la/src/../graphs/PRG_MHC_GRCh38_withIMGT/knownReferences - create a file for this BAM file and try again. at /home/admin/anaconda3/bin/HLA-LA.pl line 309

Below I am attaching samtools idxstats results for my BAM file
Kindly help me to resolve the issue
samtools_idsstats.txt

[TonyLupara]

Hi, I've corrected your file to use as extraction specifications by HLA-LA. It contains information about what reads to extract from your BAM file to use in genotyping. I marked chr6 MHC region and all unmapped reads.

Add this file into graphs/PRG_MHC_GRCh38_withIMGT/knownReferences folder.

reference_extraction.txt

Let me know if it work.

[santhoshnh]

Hii I am getting following error after adding reference_extraction.txt into graphs/PRG_MHC_GRCh38_withIMGT/knownReferences folder

Incorrect header for /home/admin/anaconda3/opt/hla-la/src/../graphs/PRG_MHC_GRCh38_withIMGT/knownReferences/reference_extraction.txt at /home/admin/anaconda3/bin/HLA-LA.pl line 255, line 1

[TonyLupara]

Use this one, it should work

reference_extraction_UNIX.txt

[santhoshnh]

Thank you. It is working. Can I know what is the difference between this file and previous one?? So that if I use different genome I can make it by my own

[TonyLupara]

The main idea is that when you edit text file in Windows it uses CR LF (Windows) line break type, I have changed it with Notepad++ to LF (Unix) break type.

[santhoshnh]

Thank you. Can I use the bam file which is obtained by mapping fasta reads to reference genome??

[TonyLupara]

You can, just make samtools idxstats results for your BAM file, extract contigID and Length, make proper extraction file with Excel for example, and switch to Unix line break type.

[chenxf611]

Hi, I mapped my pacbio HLA reads to hg38 and then run HLA-LA.pl with --longReads pacbio tag, I got same error:

Have found no compatible reference specifications in ./graphs/PRG_MHC_GRCh38_withIMGT/knownReferences - create a file for this BAM file and try again.

My question is, which reference should I use to map my reads in the first place, regular hg38 or specified HLA reference?

Thanks

Jack

[TonyLupara]

Use whatever reference you think is correct in this situation. For example I use GRCh38_full_analysis_set_plus_decoy_hla.fa for my processing, as I believe it increase mapQ quality for reads aligned to alternative contigs. Notice that reads should be mapped to reference with alt-contigs with alt-aware mapper (for example, bwa-mem) and ideally with Postprocessing. Otherwise read maps to multiple locations will have zero mapQ.
More details here:
https://github.com/lh3/bwa/blob/master/README-alt.md
https://lh3.github.io/2017/11/13/which-human-reference-genome-to-use

If you don't know how to use alt-aware mapper its better to work with hg38_analysis_set without alt-contigs. For example:
https://ftp.ncbi.nlm.nih.gov/genomes/all/GCA/000/001/405/GCA_000001405.15_GRCh38/seqs_for_alignment_pipelines.ucsc_ids/GCA_000001405.15_GRCh38_no_alt_plus_hs38d1_analysis_set.fna.gz

[AndresCongenica]

Hi, I am trying to evaluate different tools and have chosen to use NF-Core's reference file for cutting down to relevant regions. I am facing the same issue as the rest in the thread: "Have found no compatible reference specifications in..."

Kindly see my idxstats file attached and please let me know of anything else you may require.
idx_stats.txt

[AlexanderDilthey]

Hi all,

If you have freedom to choose the reference you map against, I would recommend using the standard 1000 Genomes reference: http://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome/GRCh38_full_analysis_set_plus_decoy_hla.fa

@AndresCongenica, it looks like all of the contigs in your file contain a "HLA" substring, which may indicate that all reads mapping to any of these should be extracted. However, I am not sure where any such BAM would come from - would it be based on extracting alignment records from a BAM that is based on mapping against a whole-genome reference (which would probably be fine), or based on mapping a set of whole-genome reads against a reference containing only the HLA reference contigs (which may be a process prone to attracting false-positive alignments)?

Best wishes

Alex