HomeScienceDeep Learning Maps Jumping Genes in Pr
SCIENCE

Deep Learning Maps Jumping Genes in Primate Brain Development

Researchers used multiomics data and a deep-learning model to show how transposable elements shape gene regulation during cerebellum development.

WHAT YOU NEED TO KNOW
  • 17.6% of human cerebellar candidate cis-regulatory elements overlap with transposable elements.
  • Human-specific regulatory elements showed 62% transposable element overlap compared to 5% in mammalian-conserved elements.
  • The DeepCeREvo model identified twelve transposable element subfamilies with high cell-type-specific regulatory potential.

Researchers applied single-cell multiomics data and a deep-learning model to evaluate how transposable elements shape gene regulatory networks during primate cerebellum development, according to a study published in Nature Communications. The study analyzed chromatin accessibility data from human, macaque, marmoset, and mouse cerebella alongside the deep-learning model DeepCeREvo.

The team found that 17.6% of 557,491 human cerebellar candidate cis-regulatory elements overlap with transposable elements. Relative to the overall genome background, these regulatory elements showed a 62% reduction in transposable element overlap odds. Distal elements exhibited the highest overlap at 23.8%. Relative to distal elements, exonic regulatory elements showed an odds ratio of 0.27, promoters recorded 0.67, and intronic elements recorded 0.78.

Lineage comparisons showed that human-specific regulatory elements had 62% transposable element overlap, while mammalian-conserved elements contained 5% overlap. Marmoset-specific elements showed 63% overlap and mouse-specific elements showed 34%. Regulatory elements accessible during embryonic development had 58% lower odds of containing transposable elements than adult-accessible elements.

Subfamily Recruitment

DeepCeREvo scored 500-base-pair sequence fragments across consensus sequences to identify twelve transposable element subfamilies with high regulatory potential for specific cell types. The MER130 subfamily showed regulatory activity in differentiating granule cells through sequence motifs including E-box and NFI binding sites. Subfamilies MER41B, MER49, and MER72 showed enrichment in ventricular zone neuroblasts with POU motifs, while MamTip2b targeted mature Purkinje cells and MLT1M targeted microglia.

Chromatin accessibility profiles showed concentrated activity at the 2000 to 2500 nucleotide position of the HERVL consensus sequence in differentiating granule cells, granule cell progenitors, and differentiating unipolar brush cells.

Xentir Media
Xentir Media NewsroomSource-backed AI and technology coverage, drafted by Xentir's automated editorial system under fixed human-set rules. See our editorial policy and AI usage policy.
J
Jomon · Founder & EditorFounder and editor of Xentir Media. Sets the editorial rules the newsroom system runs under, and is accountable for its corrections. About Jomon · [email protected]
The Xentir Brief
The developments worth knowing — one useful email.
Get the Brief →