Compacta

Compacta clusters contigs from de novo assembled RNA-Seq transcriptomes by grouping sequences based on shared reads to produce reduced sets of representative sequences that retain read-based relative expression information.

Key Features:

• Graph-based clustering: A graph-based algorithm clusters contigs according to the proportion of shared reads among them.
• Shared-read threshold: A definable threshold for the proportion of shared reads controls cluster formation.
• Minimum coverage parameter: A minimum coverage parameter determines which contigs are eligible for clustering.
• Preservation of expression: Representative sequences are selected to preserve relative expression levels indicated by reads.
• Variant and paralog handling: Explicit handling of immature mRNAs, spliced transcripts, allele variants, and products of closely related paralogs or gene families.
• Contig set compression: Reduces large contig collections into smaller, representative sets for downstream analysis.
• Benchmarking: Performance evaluated against other clustering algorithms using assemblies from Arabidopsis, mouse, and mango, reporting rapid results with competitive precision and recall.

Scientific Applications:

• Transcriptome compression: Compression of de novo assembled transcriptomes from RNA-Seq to reduce redundancy in contig sets.
• Contig identification: Facilitates contig identification in assemblies, including genomes lacking comprehensive annotation.
• Differential gene expression: Provides optimized representative contig sets suitable for differential gene expression analyses.
• Method benchmarking: Enables benchmarking of clustering approaches across organisms such as Arabidopsis, mouse, and mango.

Methodology:

Compacta applies a graph-based algorithm that clusters contigs by the proportion of shared reads, using configurable parameters for minimum contig coverage and the shared-read proportion to produce representative sequences that preserve read-based relative expression.