Back

Genetic architecture of telomere length in 462,675 UK Biobank whole-genome sequences

Burren, O. S.; Gaithersburg, R. D.; Deevi, S. V. V.; Wen, S.; Nag, A.; Mitchell, J.; Hu, F.; Smith, K. R.; Razdan, N.; Olsson, H.; Platt, A.; Codd, V.; Nelson, C.; Samani, N.; March, R.; Wasilewski, S.; Carss, K.; Fabre, M.; Wang, Q.; Pangalos, M.; Petrovski, S.

2023-09-18 genetic and genomic medicine
10.1101/2023.09.18.23295715 medRxiv
Show abstract

Telomeres protect the ends of chromosomes from damage, and genetic regulation of their length is associated with human disease and ageing. We developed a joint telomere length (TL) metric, combining both qPCR and whole genome sequencing (WGS) measurements across 462,675 UK Biobank participants that increased our ability to capture TL heritability by 36% (h2mean=0.058 to h2combined=0.079) and improved predictions of age. Exome-wide rare variant (minor allele frequency<0.001) and gene-level collapsing association studies identified 53 variants and 22 genes significantly associated with TL that included allelic series in ACD and RTEL1. Five of the 31 rare-variant TL associated genes (16%) were also known drivers of clonal haematopoiesis (CH), prompting somatic variant analyses. Stratifying by CH clone size, we uncovered novel gene-specific associations with TL, including lengthened telomeres in individuals with large SRSF2-mutant clones, in contrast to the progressive telomere shortening observed with increasing clonal expansions driven by other CH genes. Our findings demonstrate the impact of rare variants on TL with larger effects in genes associated with CH, a precursor of myeloid cancers and several other non-malignant human diseases. Telomere biology is likely to be an important focus for the prevention and treatment of these conditions.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"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.