SEQ-EM

SEQ-EM estimates expression levels of homologous genes from short-read RNA-seq data by using a probabilistic model that integrates multimapping reads and accounts for genetic variability.

Key Features:

• Handling Multimapping Reads: Integrates reads that map to multiple genomic locations into its probabilistic model rather than discarding them.
• Probabilistic Modeling: Uses a maximum likelihood-based probabilistic framework to estimate expression parameters from RNA-seq data.
• Accounting for Genetic Variability: Considers genetic discrepancies between the sequenced DNA and the reference genome, including single nucleotide polymorphisms (SNPs), when estimating RNA concentrations.
• Expectation-Maximization and Variant Calling: Implements an Expectation-Maximization (EM) approach for expression estimation, with the current version not incorporating SNP variant calling within the EM algorithm.

Scientific Applications:

• Precise Expression Quantification: Quantification of gene expression levels in RNA-seq experiments requiring accurate measurement of homologous or closely related genes.
• Complex Genomes and Paralogous Genes: Analysis of expression in complex genomes or gene families where multimapping reads and sequence similarity confound read assignment.
• Improved RNA-seq Sensitivity: Enhancing accuracy and statistical power of RNA-seq studies to investigate gene expression patterns and regulatory mechanisms.

Methodology:

Maps short RNA-seq reads to a reference genome, uses read counts per gene as indicators of RNA concentration, and applies a maximum likelihood probabilistic model (implemented via Expectation-Maximization) that integrates multimapping reads and accounts for genetic discrepancies between the sequenced DNA and the reference.