Conterminator

Conterminator detects and removes incorrectly labeled sequences across biological kingdoms to improve the integrity of genomic reference databases.

Key Features:

• All-against-all comparisons: Performs exhaustive all-against-all sequence comparisons to detect sequence similarity indicative of mislabeling or contamination.
• Cross-kingdom detection: Identifies and removes incorrectly labeled sequences across different biological kingdoms.
• Database contamination reporting: Reported contamination counts of 2,161,746 sequences in RefSeq, 114,035 in GenBank, and 14,148 in the NCBI non-redundant (NR) database.
• Scalability: Scales linearly with input size to enable analysis of very large datasets.
• Performance example: Processed 3.3 terabytes of sequence data in approximately 12 days on a 32-core system.
• Genome scope: Detects contamination in both draft genomes and assemblies labeled as "complete" for model organisms.

Scientific Applications:

• Database curation: Identification and removal of mislabeled sequences to maintain high-quality reference databases.
• Public repository quality control: Detection and quantification of contamination in public sequence repositories such as RefSeq, GenBank, and NR.
• Evolutionary studies: Improves reliability of evolutionary analyses by removing cross-kingdom labeling errors.
• Functional genomics: Supports functional-genomics analyses by ensuring sequences used are correctly labeled.

Methodology:

Performs exhaustive all-against-all sequence comparisons and scales linearly with input size.