Kmerator

Kmerator predicts and analyzes specific k-mers from RNA-sequencing (RNA-seq) data to enable k-mer based transcript quantification and signature extraction using a reference genome and an ENSEMBL-like transcriptome.

Key Features:

• Specific k-mer prediction: Predicts and analyzes specific k-mers (tags) derived from input sequences for gene- and transcript-level signatures.
• Contig generation: Generates contigs as sequences of consecutive k-mers overlapping by k-1 from specific k-mers.
• Jellyfish indexing: Uses Jellyfish to build two requestable indexes from the reference genome and the ENSEMBL-like transcriptome.
• K-mer decomposition and counting: Decomposes input transcript or gene sequences into k-mers and counts occurrences within both the genome and the transcriptome.
• Human gene coverage and accuracy: Produces specific k-mers for 97% of human genes and enables accurate gene expression quantification in simulated datasets.
• Suite composition: Forms part of the Kmerator Suite comprising three tools that extract, quantify, and analyze k-mer signatures, including KmerExploR.
• Metadata inference (KmerExploR): Applies gene-specific k-mer predictors to infer metadata such as library protocol, sample features, and potential contaminations.
• Advanced queries: Supports queries for known or novel biomarkers, including identification of mutations, gene fusions, and long non-coding RNAs.
• Computational efficiency: Provides a k-mer decomposition alternative to alignment-based methods to reduce computational resource requirements for large-scale RNA-seq analyses.

Scientific Applications:

• Gene expression quantification: Quantifies transcript- and gene-level expression from RNA-seq via specific k-mer counts.
• Sample metadata inference: Infers library protocol, sample features, and contaminations from RNA-seq datasets using KmerExploR.
• Biomarker discovery: Enables queries for known or novel biomarkers from k-mer signatures.
• Detection of genetic events: Identifies mutations and gene fusions from k-mer patterns.
• Long non-coding RNA analysis: Detects long non-coding RNAs using k-mer based signatures.
• Large-scale RNA-seq processing: Facilitates resource-efficient analysis of large RNA-seq cohorts using k-mer decomposition instead of alignment.

Methodology:

Uses Jellyfish to create two requestable indexes from the reference genome and the ENSEMBL-like transcriptome; decomposes input transcript or gene sequences into k-mers and counts their occurrences within genome and transcriptome; generates contigs by joining consecutive specific k-mers overlapping by k-1; predicts specific k-mers (tags) derived from input sequences.