Back

Sparse Sequencing permits accurate and efficient quantification of genome-wide cytosine modification levels

Loo, C. E.; Fowler, J. M.; Zhu, H.; Krapp, C.; Zhu, R.; Bartolomei, M.; Zhou, W.; Kohli, R. M.

2025-02-03 genomics
10.1101/2025.01.08.631958 bioRxiv
Show abstract

5-methylcytosine (5mC) and 5-hydroxymethycytosine (5hmC) play crucial roles in epigenetic gene regulation, with dynamic levels in development and disease. While high-depth, base-resolution whole genome studies offer the most detailed view of epigenetic landscapes for individual samples, many open questions are answered by surveying changes in 5mC/5hmC levels genome-wide or at specific genomic elements across larger cohorts of samples. Nonetheless, current global quantification methods, including mass spectrometry and immunochemistry, are typically limited in accessibility, throughput, or accuracy. Here, using computational down-sampling of deeply sequenced data, we first demonstrate that sequencing base-resolution libraries to shallow depth can be sufficient for highly accurate quantification. Sparse sampling (<0.24% of the genome) can precisely measure genomic 5mC/5hmC levels (error <5%), even when levels are low (<0.3%). Using a combined chemical/enzymatic workflow, we then validate that Sparse-Sequencing (Sparse-Seq) shows high accuracy and less variability than mass spectrometry, while distinctively preserving genomic context. Applying Sparse-Seq to track developing mouse brains serially revealed an earlier emergence of 5hmCpG compared to 5mCpH and uncovered previously overlooked, genomic feature-specific differences in epigenetic dynamics. This work establishes a rigorous foundation for employing Sparse-Seq as a highly accessible approach for 5mC/5hmC quantification, allowing for economical and high-throughput analysis of epigenetic landscapes.

Published in NAR Genomics and Bioinformatics (predicted rank #8) · training set

Matching journals

The top 5 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.