Approaches for sRNA Analysis of Human RNA-Seq Data: Comparison, Benchmarking

Abstract

Expression analysis of small noncoding RNA (sRNA), including microRNA, piwi-interacting RNA, small rRNA-derived RNA, and tRNA-derived small RNA, is a novel and quickly developing field. Despite a range of proposed approaches, selecting and adapting a particular pipeline for transcriptomic analysis of sRNA remains a challenge. This paper focuses on the identification of the optimal pipeline configurations for each step of human sRNA analysis, including reads trimming, filtering, mapping, transcript abundance quantification and differential expression analysis. Based on our study, we suggest the following parameters for the analysis of human sRNA in relation to categorical analyses with two groups of biosamples: (1) trimming with the lower length bound = 15 and the upper length bound = Read length − 40% Adapter length; (2) mapping on a reference genome with bowtie aligner with one mismatch allowed (-v 1 parameter); (3) filtering by mean threshold > 5; (4) analyzing differential expression with DESeq2 with adjusted p-value < 0.05 or limma with p-value < 0.05 if there is very little signal and few transcripts.

Assessing the Differential Methylation Analysis Quality for Microarray and NGS Platforms

“…To address these problems, we applied the Hobotnica metric (H-score) [20] that we previously developed to assess the quality of molecular signatures obtained by the differential analysis of two or more groups of samples with different phenotypic characteristics and validated for DGE and DM signatures. In this way, H-scores of different DM signatures may be compared, allowing the direct evaluation of the models' performance for a particular dataset by assessing the quality of phenotypes separation, delivered by a particular signature [21]. No metric has previously been developed that evaluates the DM signature's quality in the context of a particular data set (e.g., inter-and intra-group samples distances) without a list of gold-standard DMC.…”

Benchmarking of five NGS mapping tools for the reference alignment of bacterial outer membrane vesicles-associated small RNAs

“…The advancement of high-throughput sequencing technology has led to a burst of knowledge about the complexity and diversity of small RNAs (sRNAs), but has also raised new and specific bioinformatics challenges related to the analysis of sRNA-seq data ( Baldrich et al, 2019 ; Diallo and Provost, 2020 ; Bezuglov et al, 2023 ). Most of these challenges are related to the short length of different sRNAs.…”

reanalyzerGSE: tackling the everlasting lack of reproducibility and reanalyses in transcriptomics

“…Bioinformatics is up to the challenge and many statistical approaches and protocols have been popularized [6,7]. However, despite remarkable efforts, their performance may be variable and all are far from being a standard all-in-one solution, with methods being adjusted almost on a case-by-case basis [8][9][10][11][12]. In the current context of reproducibility crisis [13][14][15][16][17][18], published results are also typically incorporated in following studies, but new hypotheses would often require updating reference genome and annotation and fine-tuned reanalyses, which are rare.…”