Cellular identity can be inferred by multi-omic approach, study in Nature claims

Researchers in Spain and Luxembourg recently proposed a computational method for reconstructing gene regulatory networks (GRNs) from gene expression data to infer cellular identity.

The case was made in a study published in the Aug. 24 edition of Nature. 

The study was conducted by Antonio del Sol Mesa, a full professor and chief scientist in bioinformatics at the University of Luxembourg, and Sascha Jung, a post-doctoral researcher in Bizkaia Technology Park's computational biology lab in Derio, Spain.

"Here, we present a computational method, Moni [Multi-omics network inference], that systematically integrates epigenetics, transcriptomics, and protein–protein interactions to reconstruct GRNs among core transcription factors and their co-factors governing cell identity," the study's abstract said. "We applied Moni to 57 datasets of human cell types and lines and demonstrate that it can accurately infer GRNs, thereby outperforming state-of-the-art methods."

Moni integrates multiple-omics datasets to "unveil" GRN's controlling cell identity, the study said.

"Although, we demonstrated the performance of Moni using human cell types and lines in this study, it is applicable to samples from other species," the study said.

Moni would be "especially" useful in studying mouse cells and publicly available query datasets obtained from the International Human Epigenome Consortium, the study said.

"Thus, we expect Moni to be a valuable tool for obtaining mechanistic insights into key transcriptional regulators in processes such as cell conversion and disease phenotypes," the study said.

The study describes Moni as "a computational approach that systematically integrates histone modification, chromatin accessibility, and transcriptomics data with a global atlas of transcription factor [TF]-binding events, enhancer–promoter interactions, and protein–protein interactions across diverse cell types and lines."

Cellular phenotypes are stable gene expression profiles maintained by TFs to determine cell identity and, combined with other factors, "form a regulatory core network, which is shaped by the epigenetic landscape," the study said.

Scientists have long relied on mechanistic cell models to characterize and predict how cells will identify and, ultimately, help form a full-fledged organism.

"In the past decades, a wealth of computational methods has been developed that aim at identifying regulatory interactions between genes," the introduction to Mesa and Jung's study said. "However, these methods require tremendous amounts of transcriptomics data and cannot provide information about the active regulatory regions, an issue that even new technologies such as single-cell profiling do not mitigate."

Computational approaches have as long been proposed to reconstruct gene regulatory networks, or GRNs, from what it called "gene expression data," or how a gene will manifest itself as part of a final product.

The difficulty with these computational approaches is that gene regulatory processes are often too complex to predict how a gene will express based only on the transcriptome, the messenger RNA (mRNA).

To address those shortcomings, Mesa and Jung's study proposes that Moni maps phenotype-specific core regulatory network, "including regulation at distal enhancer regions and the cooperatively of TFs in the regulation of target genes."

"With the steady increase in epigenetic profiling, we expect Moni to be of general utility for the molecular characterization of cellular phenotypes and to aid in the identification of key regulators," the study said.