A Harvard University-led team describes a pipeline for processing and analyzing single-cell transcriptome data generated with the help of droplet microfluidics approaches. The method, known as dropEst, is designed for use with several single-cell RNA sequencing protocols and alignment approaches, the researchers note, and includes steps to assess library and cell quality, while correcting for confounding factors related to unique molecular identifier coverage, sequencing errors, and other issues that may arise. "[W]e hope that the developed pipeline will facilitate analysis of droplet-based single-cell RNA-seq data," they write, "providing helpful diagnostics."
Chinese researchers introduce DeepCRISPR, a computational strategy for designing optimal single guide RNAs — while taking into account the target sequence and epigenetic factors that may impact their effectiveness — in CRISPR-based gene editing experiments. The method brings together on- and off-target sgRNA predictions into a deep learning framework that compared favorably with alternative in silico tools that are typically used for doing such predictions. "We believe that future insights from the deep learning community as well as the data accumulation in the genome editing community will lead to enhancements of DeepCRISPR and CRISPR-based gene editing analysis generally," the authors say.
Finally, a University of Pennsylvania team presents an approach for identifying biologically relevant microbial taxa from 16S ribosomal RNA gene amplicon sequences or other types of targeted microbial gene sequencing. This approach — called the "Hidden Markov Model-based Ultra-Fast OTU tool," or HmmUFOtu — is designed to process such amplicon sequence data and makes taxonomic classifications and sequence inferences, respectively, informed by phylogenetic positions and consensus sequence information. The researchers demonstrated the feasibility of the approach using simulated and real 16S sequence datasets, including data from the Human Microbiome Project.