Researchers from the Children's Hospital of Philadelphia and the University of Pennsylvania outline a computational method for predicting gene expression based on single-cell DNA methylation profiles. The approach — known as "methylome association by predictive linkage to expression," or MAPLE — deciphers DNA methylation-gene activity relationships with the help of available gene and cell features, along with a supervised learning strategy, the team explains. The authors validated this approach with several single-cell datasets produced with distinct single-cell methylation and transcriptome profiling protocols, before applying it to almost 3,400 individual methylation-profiled mouse neurons and other single-cell datasets. "With the rapid accumulation of single-cell epigenomics data, MAPLE provides a general framework for integrating such data with transcriptome data," the authors note.
A team from the J. Craig Venter Institute, the University of California, Los Angeles, and UC-San Diego describes oral microbe community features found in children with dental caries, commonly known as cavities. By comparing de novo metagenomic sequence assemblies from 23 children with caries and two dozen without, along with related host cytokine and chemokine immune features, the researchers identified 150 bacterial species, along with viruses, bacterial representatives, and host immune markers that appeared to be over-represented in the mouths of children with or without caries. "Overall," they write, "this study illustrated the oral microbiome at an unprecedented resolution, and contributed several leads for further study that will increase the understanding of caries pathogenesis and guide therapeutic development."
Finally, investigators from the Chinese Academy of Sciences, UT Southwestern Medical Center, and Boston University present a computational tool known as MAnorm2 — designed for quantitative comparisons between chromatin immunoprecipitation-sequencing (ChIP-seq) datasets — that reportedly compares favorably with other available analytical strategies for comparing sets of ChIP-seq-profiled samples. "MAnorm2 employs a hierarchical strategy for normalization of ChIP-seq data and assesses within-group variability of ChIP-seq signals based on an empirical Bayes framework," the team explains, noting that the strategy "allows for abundant differential ChIP-seq signals between groups of samples as well as very different global within-group variability between groups."