SeqWho

SeqWho classifies FASTQ(A) sequencing files by species of origin and sequencing protocol to enable accurate metadata assignment for downstream genomic analyses of public repositories such as SRA, EGA, and ENCODE.

Key Features:

• Alignment-Free Algorithm: Uses k-mer frequencies and repeat sequence identities to characterize sequencing files without sequence alignment.
• Random Forest Classifier: Employs a Random Forest trained on biases in k-mer distributions and repeat sequences across species and sequencing technologies.
• High Accuracy Rates: Distinguishes human and mouse samples with 98.32% accuracy, identifies nine sequencing technologies with 97.86% accuracy, and jointly classifies species and technology with 96.38% accuracy.
• Rapid Processing: Optimized for rapid assessment of FASTQ(A) files to support high-throughput analyses.

Scientific Applications:

• Metadata verification: Flags misannotated or missing metadata in sequencing files from repositories such as SRA, EGA, and ENCODE.
• Species identification: Verifies species of origin (e.g., human versus mouse) in FASTQ(A) datasets to prevent biological misinterpretation.
• Protocol/technology identification: Identifies sequencing protocols and technologies (nine classes reported) to harmonize analyses across datasets.

Methodology:

Alignment-free analysis of k-mer frequencies and repeat sequence identities combined with a Random Forest classifier trained on biases in k-mer distributions and repeat sequences across species and sequencing technologies.