A heavy-tailed model for analyzing miRNA-seq raw read counts

Area of Science:

• Bioinformatics
• Statistical Genetics
• Genomics

Background:

• MicroRNA sequencing (miRNA-seq) data exhibit high skewness and zero counts, posing challenges for standard statistical models.
• Existing models often fail to adequately capture the heterogeneity and extreme values in miRNA-seq raw read counts.

Purpose of the Study:

• To propose a novel statistical approach for analyzing highly skewed miRNA-seq raw read count data.
• To introduce discrete stable distributions as a superior alternative to traditional models for miRNA-seq data.
• To offer parameters of the discrete stable distribution as a new target for differential expression analysis.

Main Methods:

• Development and application of a heavy-tailed model based on discrete stable distributions.
• Creation of an R package for computing and estimating discrete stable distributions.
• Comparative analysis of model goodness-of-fit against Poisson and negative binomial distributions using real-world datasets.

Main Results:

• Discrete stable distributions provide a significantly better fit to miRNA-seq raw counts compared to Poisson and negative binomial distributions.
• The proposed model effectively captures data heterogeneity and extreme values prevalent in miRNA-seq datasets.
• Application to Norwegian Women and Cancer Study (NOWAC) and Cancer Genome Atlas (TCGA) data demonstrates superior performance.

Conclusions:

• Discrete stable distributions offer a more accurate and robust method for modeling miRNA-seq raw read count data.
• This novel approach has the potential to enhance the understanding of underlying biological processes through improved statistical analysis.
• The developed R package facilitates the practical implementation of discrete stable distributions in miRNA-seq research.