Decoding enhancer complexity with machine learning and high-throughput discovery

Abstract

Enhancers are genomic DNA elements controlling spatiotemporal gene expression. Their flexible organization and functional redundancies make deciphering their sequence-function relationships challenging. This article provides an overview of the current understanding of enhancer organization and evolution, with an emphasis on factors that influence these relationships. Technological advancements, particularly in machine learning and synthetic biology, are discussed in light of how they provide new ways to understand this complexity. Exciting opportunities lie ahead as we continue to unravel the intricacies of enhancer function.

BRAIN-MAGNET: A novel functional genomics atlas coupled with convolutional neural networks facilitates clinical interpretation of disease relevant variants in non-coding regulatory elements

“…Here we used ChIP-STARR-seq to build an extensive atlas of NCREs functionally active in NSCs. Besides informing on the biological mechanisms regulating gene expression in NSCs and on NCREs distinguishing sequence characteristics, we provide evidence that a CNN model trained on the experimental data allows to predict NCRE activity solely based on NCRE nucleotide composition, similar to what other recent MPRA studies have found (Bravo Gonzalez-Blas et al, 2024; de Almeida et al, 2022; Smith et al, 2023). BRAIN-MAGNET can be used to predict the functional effects of genomic variants overlapping with NCREs, and such predictions occur with high confidence as testified by our functional validation.…”

“…First, MPRAs measure NCRE activity in an episomal context, outside of the natural chromatin environment, hence, results might not always reflect the endogenous NCRE activity. Notwithstanding this potential limitation, previous work has already extensively shown that multiple MPRA findings can be reproduced when altering NCREs at the endogenous locus (Barakat et al ., 2018; Lim et al, 2024; Smith et al ., 2023). To minimize potentially confounding effects, we generated ChIP-STARR-seq plasmid libraries from chromatin marked in NSCs with histone modifications associated with active NCREs.…”

Targeting circadian transcriptional programs through a cis-regulatory mechanism in triple negative breast cancer

“…According to the “TF collective” model of enhancer grammar, numbers of each TF-binding motif in an enhancer exhibit certain degrees of flexibility for its activity. 28 We thus excluded consideration of the numbers of each motif and instead focused solely on their presence or absence. The motif count matrices thus can be collapsed into binary-entry matrices that indicate if the motif exists (“1”) in a peak or not (“0”).…”

Gene replacement therapies for inherited disorders of neurotransmission: Current progress in succinic semialdehyde dehydrogenase deficiency

“…However, traditional promoter studies often exploit heterologous recombinant cell‐based systems such as fibroblasts for excellent transfection efficiency and reporter assay compatibility, but the cellular signaling and genetic contents are often biased or irrelevant for physiologic transgene expression. A fundamentally different approach to designing a promoter would involve an unbiased assessment of transgene expression in relevant cell types (e.g., single cell RNA‐seq), combined with bioinformatics 72 and machine learning approach 73–75 to “mine” patterns of DNA sequences correlating all genes' expression profiles across cell types (Figure 3). Ideally, these regulatory elements or promoters should drive a given gene's functional expression resembling its endogenous profile across developmental stages.…”