Back to Search Start Over

P055 GFE ACT: A community resource for calling HLA allele names from consensus sequences with gene feature enumeration notation.

Authors :
Halagan, Michael S.
Wang, Wei
Bashyal, Pradeep
Heuer, Michael
Mack, Steven J.
Maiers, Martin
Source :
Human Immunology. Sep2017 Supplement, Vol. 78, p94-94. 1p.
Publication Year :
2017

Abstract

Aim A publicly available service for assigning HLA allele names to consensus sequences would be a valuable community resource. Our aim was to use existing technologies to develop a publicly available, graph-based HLA allele calling service. Methods Gene feature enumeration (GFE) notation has been developed as a way of describing sequence variation outside of current HLA nomenclature. We generated GFE notation for every sequence in the IPD-IMGT/HLA Database, and loaded the results into a neo4j graph database ( neo4j.b12x.org ), which includes HLA, GFE and sequence feature nodes. We developed an allele calling tool (ACT) that leverages this graph to return an HLA allele call and GFE notation from submitted consensus sequence. ACT uses the GFE service ( gfe.b12x.org ) to convert the consensus sequences to GFE notation, and then searches the graph database for HLA alleles that share the most features with the generated GFE. A RESTful service interface makes ACT easy to use and allows for cross-platform compatibility ( act.b12x.org ). We used 30,000 Be the Match® donors typed at high resolution and with consensus sequences available for HLA-A, B, and C to test ACT. We compared the HLA calls ACT produced to the allele names reported by the typing labs. Comparisons were made with and without mapping alleles to their G group equivalent. Results When mapping alleles to their respective G group equivalent, the reported HLA typing for all 30,000 donors matched the ACT typing to the 2nd field with no ambiguity. Without mapping alleles to their G groups, the number of exact matches to the 4th field at HLA-A, B and C was 17,142 (57%), 16,540 (55%) and 10,024 (33%), respectively. Using G groups, ambiguity at the 4th field was reduced substantially; the number of exact matches increased to 29,914 (99.7%), 29,733 (99.1%) and 29,985 (99.9%) for HLA-A, B, and C, respectively. In 75 instances, ACT called alleles at a higher resolution than reported by the typing lab. Conclusions HLA class I allele names can be accurately called from consensus sequences using our GFE-based ACT. This service will allow anyone to easily convert HLA class I consensus sequences into the corresponding GFE notation and HLA allele name. Class II HLA and KIR are already represented in the graph, and we plan to extend ACT functions to these loci. [ABSTRACT FROM AUTHOR]

Details

Language :
English
ISSN :
01988859
Volume :
78
Database :
Academic Search Index
Journal :
Human Immunology
Publication Type :
Academic Journal
Accession number :
124999766
Full Text :
https://doi.org/10.1016/j.humimm.2017.06.115