A Chinese team takes a look at the "hidden" human proteome — functional proteins encoded by sequences previously thought to be non-coding RNAs. After using RNA sequencing and full-length translating messenger RNA-seq to find thousands of long non-coding RNAs (lncRNAs) bound to ribosomes with active translation elongation in human cell lines, the researchers turned to shotgun proteomics to identify more than 300 proteins encoded by lncRNAs. They went on to validate and characterize a subset of those proteins and peptides with approaches such as immunoblotting, multiple reaction monitoring, parallel reaction monitoring, and translation efficiency assays. "[O]ur experimental validation of the hidden human proteome encoded by the 'non-coding' RNAs will move the field forward by distinguishing a remarkable proportion of the lncRNAs that function with their encoded proteins," the authors write, noting that the findings "will fundamentally improve our understanding on disease models."
French researchers present an algorithm for gleaning expression signatures from single-cell RNA sequence data. The qualitative, high-throughput software package — called Single-Cell Signature Explorer — tracks down differentially-expressed gene sets with signature scoring, collation, and combination detector tools that are coupled with available visualization tools, the team explains. In their study, for example, the authors applied Single-Cell Signature Explorer to single-cell RNA-seq data for thousands of B cell lymphocyte cells from peripheral blood mononuclear cell samples, melanoma tumor cells, lung cancer tumor cells, and individual cells from adult testis samples. "By quickly delineating multi-gene features such as cell lineage or metabolic pathways with Single-Cell Signature Explorer in lung tumors and normal human blood and testis," they report, "we showed that gene set-based signatures outperform single genes and provide a straightforward visualization of the sample's hallmarks."
Finally, a team from China describes a statistical strategy for discerning between cancer driver genes and genes prone to passenger mutations. The researchers' "weighted iterative zero-truncated negative-binomial regression," or WITER, uses an unsupervised, three-tiered structure to search for potential driver genes that are routinely impacted by somatic single nucleotide changes or small insertions and deletions, while weeding out background mutations. When they used WITER to assess tumors from 26 cancer types, the authors narrowed in on more than 200 apparent cancer driver genes, using in silico validation to verify more than three-quarters of the predicted drivers. Based on driver gene profiles across 11 cancer types, they report, WITER also appeared to compare favorably with other available methods for finding cancer driver genes.