Back

Validated machine learning algorithm with sub-clonal sensitivity reveals widespread pan-cancer human leukocyte antigen loss of heterozygosity

Pyke, R. M.; Mellacheruvu, D.; Dea, S.; Abbott, C. W.; Zhang, S. V.; McDaniel, L.; Levy, E.; Bartha, G.; West, J.; Snyder, M. M.; Chen, R.; Boyle, S. M.

2021-06-28 bioinformatics
10.1101/2021.05.20.445052 bioRxiv
Show abstract

Human leukocyte antigen loss of heterozygosity (HLA LOH) allows cancer cells to escape immune recognition by deleting HLA alleles, causing the suppressed presentation of tumor neoantigens that would otherwise bind to them. Despite its importance in immunotherapy response, few methods exist to detect HLA LOH, and their accuracy is not well understood. Here, we develop DASH (Deletion of Allele-Specific HLAs), a novel machine learning-based algorithm to detect HLA LOH from paired tumor-normal sequencing data. Through validation with cell line mixtures and patient-specific digital PCR, we demonstrate increased sensitivity compared to previously published tools and pave the way for clinical utility. Using DASH on 611 patients across 15 tumor types, we found that 18% of patients had HLA LOH. Moreover, we show inflated HLA LOH rates compared to genome-wide LOH and correlations between CD274 (PD-L1) expression and MSI status, suggesting the HLA LOH is a key immune resistance strategy.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

1
Nature Communications
○ 5641 papers in training set
Top 8%
18.6%
Similar papers in this journal
2
Nature Biotechnology
○ 172 papers in training set
Top 0.3%
7.9%
Similar papers in this journal
3
Genome Biology
○ 637 papers in training set
Top 3%
4.3%
Similar papers in this journal
4
Genome Medicine
○ 183 papers in training set
Top 0.8%
4.3%
Similar papers in this journal
5
eLife
● 5828 papers in training set
Top 30%
4.1%
6
The American Journal of Human Genetics
○ 234 papers in training set
Top 1%
4.1%
Similar papers in this journal
7
Cancer Research Communications
● 51 papers in training set
Top 0.3%
3.4%
8
Communications Biology
○ 993 papers in training set
Top 5%
3.2%
9
Cell Genomics
○ 172 papers in training set
Top 1%
3.2%
50% of probability mass above
10
Briefings in Bioinformatics
◐ 354 papers in training set
Top 3%
2.8%
11
Science Immunology
● 88 papers in training set
Top 1%
2.8%
12
Nucleic Acids Research
◐ 1281 papers in training set
Top 6%
2.8%
13
Nature Methods
○ 385 papers in training set
Top 3%
2.8%
14
Cell Reports Methods
○ 165 papers in training set
Top 0.9%
2.4%
15
Bioinformatics
◐ 1204 papers in training set
Top 6%
2.4%
16
Scientific Reports
○ 3612 papers in training set
Top 45%
2.1%
17
Cell Reports Medicine
○ 153 papers in training set
Top 2%
1.9%
18
iScience
○ 1154 papers in training set
Top 19%
1.7%
19
Genome Research
● 468 papers in training set
Top 4%
1.5%
20
Nature Genetics
○ 286 papers in training set
Top 4%
1.1%
21
Science Advances
● 1243 papers in training set
Top 25%
1.1%
22
Cell Systems
○ 201 papers in training set
Top 3%
1.1%
23
Cancer Research
● 130 papers in training set
Top 2%
1.1%
24
NAR Genomics and Bioinformatics
◐ 242 papers in training set
Top 4%
1.0%
25
Proceedings of the National Academy of Sciences
● 2444 papers in training set
Top 37%
1.0%
26
Cell Reports
○ 1498 papers in training set
Top 26%
0.8%
27
PLOS Computational Biology
● 1863 papers in training set
Top 19%
0.8%
28
Nature Machine Intelligence
○ 70 papers in training set
Top 3%
0.6%
29
BMC Genomics
○ 406 papers in training set
Top 9%
0.6%
30
BMC Medical Genomics
○ 50 papers in training set
Top 1%
0.6%

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.