Error in generating HLA reference panel

[danielanach]

Hi there,

I installed QUILT using the "github" installation instructions, and used the most recent release (1.0.1). The installation ran smoothly and all the all the tests passed.

For the analysis I am running, I needed a slightly different version of IGMT HLA alleles database (3.43 instead of the 3.39 in the pre-made reference material), so I followed the instructions provided here to generate new reference material: https://github.com/rwdavies/QUILT/blob/master/example/QUILT_hla_reference_panel_construction.Md

To make sure it ran as expected I just used the exact instructions listed, including using the 3.39 IGMT reference material to start.

However I keep running into an error in the make_and_save_hla_full_alleles_filled_in() function.

I re-downloaded and checked all the downloaded files, but am a bit stumped.

Attached is the error message :

[2021-12-10 12:57:01] Running QUILT_HLA_prepare_reference(outputdir = /cellar/users/dnachman/programs/QUILT2/references, 
nGen = 100, hla_gene_region_file = /cellar/users/dnachman/programs/QUILT2/QUILT/hla_ancillary_files/hlagenes.txt,
hla_types_panel = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/20181129_HLA_types_full_1000_Genomes_Project_panel.txt,
ipd_igmt_alignments_zip_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/Alignments_Rel_3390.zip,
quilt_hla_supplementary_info_file=/cellar/users/dnachman/programs/QUILT2/QUILT/hla_ancillary_files/quilt_hla_supplementary_info.txt,
full_regionStart = 25587319, full_regionEnd = 33629686, buffer = 500000, 
region_exclude_file = /cellar/users/dnachman/programs/QUILT2/QUILT/hla_ancillary_files/hlagenes.txt, 
genetic_map_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/CEU/CEU-chr6-final.b38.txt.gz,
reference_haplotype_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/oneKG.hap.gz, 
reference_legend_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/oneKG.legend.gz, 
reference_sample_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/oneKG.samples,
reference_exclude_samplelist_file = /cellar/users/dnachman/programs/QUILT2/QUILT_HLA/exclude_ref_samples_for_testing.txt,
reference_exclude_samples_for_initial_phasing = FALSE, 
all_hla_regions = c("A", "B", "C", "DMA", "DMB", "DOA", "DOB", "DPA1", "DPA2", "DPB1", "DPB2", "DQA1", "DQA2", "DQB1", "DRA", "DRB1", "DRB3", "DRB4", "DRB5", "E", "F", "G", "HFE", "H", "J", "K", "L", "MICA", "MICB", "N", "P", "S", "TAP1", "TAP2", "T", "U", "V", "W", "Y"),
hla_regions_to_prepare = c("A", "B", "C", "DQB1", "DRB1"), chr = chr6, minRate = 0.1, nCores = 1)
sending incremental file list

sent 81 bytes  received 12 bytes  186.00 bytes/sec
total size is 5,691,094  speedup is 61,194.56
Archive:  Alignments_Rel_3390.zip
[2021-12-10 12:57:01] Begin making HLA all alleles kmers file
[2021-12-10 12:57:01] Processing file 1 out of 39
[2021-12-10 12:57:50] Processing file 4 out of 39
[2021-12-10 12:57:54] Processing file 8 out of 39
[2021-12-10 12:58:48] Processing file 12 out of 39
[2021-12-10 12:59:13] Processing file 16 out of 39
[2021-12-10 13:02:54] Processing file 20 out of 39
[2021-12-10 13:03:00] Processing file 24 out of 39
[2021-12-10 13:03:05] Processing file 28 out of 39
[2021-12-10 13:05:12] Processing file 32 out of 39
[2021-12-10 13:05:20] Processing file 36 out of 39
[2021-12-10 13:05:58] Done making HLA all alleles kmers file
[2021-12-10 13:05:58] Begin making HLA snp format alleles file
[2021-12-10 13:05:58] Working on region:A
[2021-12-10 13:07:18] Working on region:B
[2021-12-10 13:12:12] Working on region:C
[2021-12-10 13:16:18] Working on region:DQB1
[2021-12-10 13:16:45] Working on region:DRB1
[2021-12-10 13:18:52] Done making HLA snp format alleles file
[2021-12-10 13:18:52] Begin making HLA full alleles filled in file
[2021-12-10 13:18:52] Processing HLA-A
[2021-12-10 13:21:42] Processing HLA-B
Error in colSums(allelemat[1:5, ]) :
  'x' must be an array of at least two dimensions
Calls: QUILT_HLA_prepare_reference ... make_and_save_hla_full_alleles_filled_in -> colSums
Execution halted`

Let me know if I am missing something, I will keep de-bugging but thought I would post this in case others also encounter the same problem. Thanks!

[rwdavies]

Hi,

Sorry for my slow reply, my daughter is off sick this week from nursery, so I have limited time to work.

Thanks for the message, I'll investigate. I'm going to try re-running on my machine to see if I get the same error with the latest version of the repo. If not I'll take a deeper look into the above error log. I'll also try with the newer IGMT file as well to see if that works on my machine without issue.

Thanks Robbie

[danielanach]

Hi Robbie,

No worries at all, apologies for adding to your plate, hope your daughter gets well soon!

Best, Daniela

[rwdavies]

So the current master repo version worked for me on clean directories, can you please re-try the following (if you're on a Linux OS). I can't quite remember but the tools to build the reference panel might have check points that get thrown off if you re-run changing some of the files partway through

1) Modify the following paths to new directories on your local environment in the file example/QUILT_hla_reference_panel_construction.Md

inputs_dir=/data/smew1/rdavies/quilt_hla_2021_12_15/
test_dir=/data/smew1/rdavies/quilt_hla_2021_12_15/
reference_package_dir=/data/smew1/rdavies/quilt_hla_reference_panel_files_2021_12_15/

2) Run

bash example/run_example.sh example/QUILT_hla_reference_panel_construction.Md 

Here's the log I get when I do that on my machine log_hla_2021_12_15.txt

[danielanach]

Thank you for the suggestion, starting from scratch resolved the error when working with the 3.39 HLA versions and I was able to successfully construct a reference panel. Indeed I must have re-run with the same directories after the initial failed attempt and that must have corrupted something.

I did have to change a couple more lines in the file example/QUILT_hla_reference_panel_construction.Md that relate to the location of the QUILT installation:

R -f ~/proj/QUILT/scripts/make_b38_recomb_map.R --args ${inputs_dir} CEU 6

cd ~/proj/QUILT/ ## change to the directory where you've cloned QUILT

cd ~/proj/QUILT/

After these changes my log looked identical to yours and QUILT ran successfully on the test BAM files with the same predicted alleles.

However, when I start with clean directories and swap out the alignment file with a new version Alignments_Rel_3430.zip

I am still getting the same error in the reference construction on HLA-B as the original issue log. Not sure why yet, but at least now I can produce one successful reference package with the older HLA version!

[rwdavies]

OK thanks, great. I had some weird issues downloading data last week when running the script but looks OK now, I'll re-start and hope it captures the same error that I can then debug.

[danielanach]

It looks like the version 1.0.2 that you released resolves my issue, I was able to run on 3.43 without problems so I will close this issue. Thank you!