When silent mutations speak up: Fine granularity in linkage disequilibrium

Abstract

Aim Silent mutations and mutations in non-coding regions are traditionally considered clinically irrelevant and ignored in clinical testing. Are we missing any valuable information in the data we discard? Methods Sequences obtained by standard SBT methods are reanalysed with BLAST and ad hoc sequence databases to study sequence segments typically overlooked. Results Examples of some observations made: C∗03:03:01 goes with B∗15:01:01:01, B∗55:01:01 and B∗07:02:01, but C∗03:03:04 goes with B∗44:03:01. C∗03:04:01:01 goes preferentially with B∗40:01:02 and less frequently with B∗15:01:01:01, while C∗03:04:02 goes commonly with B∗15:10:01 and with B∗53:01:01, which is also seen with the first allele. C∗05:01:01:02 goes with B∗44:02:01:01 and C∗05:01:01:01 goes with B∗18:01:01:01. C∗06:02:01:01 goes with, B∗13:02:01, B∗57:01:01, B∗37:01:01 and B∗47:01:01:01; C∗06:02:01:02 goes with B∗50:01:01 and B∗50:02; and C∗06:02:01:03 goes with B∗45:01. C∗07:02:01:03 goes with B∗07:02 and C∗07:02:01:01 goes with a variety of B alleles including B27:05, B∗39:01 and B∗38:02. C∗15:05:02 goes with both B∗07:05 and B∗07:06, two alleles that are themselves difficult to tell apart; and C∗15:05:01 goes with B∗73:01 [Full results to be presented at ASHI 2015.] Conclusions Most of the time allele variants with silent mutations for a common allele are very rare or even poorly documented, but from time to time we see a common allele present in 2 silent-mutation forms. (Here we include intronic mutations with silent mutations.) Typically these variants are ignored and alleles are reported only with the first 2 fields or first 4 digits because only the protein product, and not the DNA sequence, is relevant in a clinical setting. When we look at linkage disequilibrium between HLA loci we find however that these silent-mutation variants of common alleles sometimes have strikingly different associations with alleles at other loci. Genotype analysis and linkage disequilibrium analysis should be based on information provided by the full DNA sequence, including silent mutations and intronic mutations, and not just on the coding sequence that defines a protein with a unique sequence of amino acids. Next generation sequencing will optimize this process.