DeconSeq

DeconSeq identifies and removes contaminant DNA sequences from metagenomic datasets generated by high-throughput sequencing to improve accuracy of downstream genome assembly and microbial diversity analyses.

Key Features:

• Automated detection and removal: Rapidly identifies and removes contaminant sequences from longer-read datasets (mean read length of 150 bp).
• Host contamination screening: Detects contamination from human and other host sequences in metagenomic data.
• Alignment technology: Uses a modified version of the BWA-SW aligner to process longer reads for contaminant detection.
• Classification of contaminants: Categorizes potential contaminant sequences and classifies alignment results.
• Redundant-hit elimination: Removes redundant hits by selecting higher similarity to non-contaminant genomes.
• Visualization of results: Produces graphical visualizations of alignment results and contaminant classifications.
• Empirical detection in published datasets: Identified possible human DNA contamination in 145 (72%) of 202 previously published microbial and viral metagenomes, with some datasets containing up to 64% contaminating sequences.
• Quality assurance indicators: Reveals whether a sequencing experiment succeeded, whether the correct sample was sequenced, and whether contamination from DNA preparation or host is present.

Scientific Applications:

• Metagenomic quality control: Screening and removal of contaminant sequences to improve data quality for downstream analyses.
• Genome assembly improvement: Removing contaminant reads to reduce misassembly of sequence contigs.
• Microbial diversity studies: Ensuring accurate assessment of microbial community composition by eliminating non-target DNA.
• Virology and microbiology research: Detecting host-derived contamination in viral and microbial metagenomes.
• Retrospective dataset screening: Screening published metagenomic datasets for human or host contamination.

Methodology:

Aligns reads using a modified BWA-SW, automatically identifies and removes contaminant sequences from longer-read datasets (mean read length of 150 bp), categorizes potential contaminants, eliminates redundant hits by choosing higher similarity to non-contaminant genomes, and generates graphical visualizations of alignments and classifications.