Susanne Bornelöv "sblab"

bioinformatics, (epi)genomics, machine learning, piRNAs, codon usage

Chromatin and epigenetics

Histone posttranslational modifications provide a critical layer of gene regulation and genome organisation. During my PhD, I noticed that the histone modification H3K36me3 correlates with the inclusion level of cassette exons. Building on this finding, I developed a rule-based machine learning model able to accurately predicts exon inclusion levels from histone modification patterns (Enroth S, Bornelöv S et al., 2012). Importantly, this suggested that splicing occurs co-transcriptionally at chromatin.

Although many histone modifications are categorised as "activating" or "repressive" marks, this notion is often based on correlation and may lack evidence of causality. To test the histone code hypothesis, I compared the occurence of histone modification and transcription factor upstream of transcription start sites (TSS) of genes that are either unidirectionally or bidirectionally transcribed. This revealed that upstream "activating" H3K4me3, H3K9ac, and H3K27ac marks reflected the level of bidirectional transcription, suggesting that they are deposited as a consequence of transcriptional activity (Bornelöv et al., 2015). Unexpectedly, the same study identified distinct CTCF and RAD21 peaks 60-80 nt upstream of unidirectional genes, providing early evidence that unidirectional transcription is a consequence of CTCF binding (Bornelöv et al., 2015).

Our current efforts in the epigenetics area include exploring the connection to the piRNA pathway and how this non-coding RNA pathway is involved in establishing and reading chromatin marks.