Modeling bias and variation in the stochastic processes of small RNA sequencing

Abstract

The use of RNA-seq as the preferred method for the discovery and validation of small RNA biomarkers has been hindered by high quantitative variability and biased sequence counts. In this paper we develop a statistical model for sequence counts that accounts for ligase bias and stochastic variation in sequence counts. This model implies a linear quadratic relation between the mean and variance of sequence counts. Using a large number of sequencing datasets, we demonstrate how one can use the generalized additive models for location, scale and shape (GAMLSS) distributional regression framework to calculate and apply empirical correction factors for ligase bias. Bias correction could remove more than 40% of the bias for miRNAs. Empirical bias correction factors appear to be nearly constant over at least one and up to four orders of magnitude of total RNA input and independent of sample composition. Using synthetic mixes of known composition, we show that the GAMLSS approach can analyze differential expression with greater accuracy, higher sensitivity and specificity than six existing algorithms (DESeq2, edgeR, EBSeq, limma, DSS, voom) for the analysis of small RNA-seq data.

Analysis and correction of compositional bias in sparse sequencing count data

“…While fully joint parameter inference algorithms will certainly be more accurate, they are unwieldy and computationally intensive with large scale datasets boasting a large number of features with high sparsity. A case in point is the GAMLSS methodology [58], which improved over our pipeline (Wrench normalization coupled with edgeR differential abundance analysis) in a small scale equimolar miRNA benchmarking dataset (Additional file 1: Figure S23), but could not run to completion even in the simplest of our metagenomic datasets, the mouse gut microbiome. Second, our simulation results indicate that the performance of Wrench stabilizes by 10 − 20 samples per group depending on sample depth and the fraction of features that change across conditions.…”

Study protocol: rationale and design of the community-based prospective cohort study of kidney function and diabetes in rural New Mexico, the COMPASS study

“…Researching microRNA biomarkers in the COMPASS population which is at extremely high risk of diabetic CKD, not only addresses an obvious research disparity but may inform future studies about the utility of these biomarkers in the general population. In fact the COMPASS project has already generated significant methodological advances in the statistical techniques for the analysis of microRNA data [ 17 ]. These techniques not only address the variability [ 29 , 30 ] and bias [ 31 – 33 ] in the short RNA sequencing measurements, but are able to estimate group differences in expression with high accuracy and precision.…”

“…Individual library concentrations are measured using the NEBNext Library Quant Kit for Illumina (New England Biolabs, Ipswich MA) and adjusted to a final pooled concertation of 2 nM and run on NextSeq sequencer (Illumina, San Diego CA). For a detailed overview of the steps involved in the construction of microRNA libraries see the supplementary methods of our short RNA sequencing analysis companion paper [ 17 ].…”

“…MicroRNA measurements (counts of sequences of distinct microRNAs) will be analyzed by our recently described Linear Quadratic Normal (LQNO) GAMLSS model [ 17 ]. This LQNO model is a statistical regression framework for the analyses of short RNA sequencing counts that takes into consideration the statistical processes underlying this complex measurement.…”

Role of microRNAs in Renal Parenchymal Diseases—A New Dimension

“…Moreover, because the biases vary depending on the library preparation protocol, comparing data generated by two different sample prep methods is difficult [101,102]. Although attempts have been made to alleviate bias in sRNAseq through modifications in library preparation methods [100,103,104] or to compensate for it during data analysis [105], it remains a significant issue that will need to be addressed in the future.…”