DiTASiC

DiTASiC quantifies taxa abundances and performs differential abundance analysis in metagenomic samples while correcting for sequence similarity to enable strain-level and sub-strain resolution.

Key Features:

• Shared-read resolution: Uses a generalized linear model to resolve reads shared between highly similar genomes and reduce ambiguity in read assignment.
• Uncertainty integration: Incorporates abundance estimation uncertainties into downstream differential abundance testing.
• Statistical framework: Constructs a framework that accounts for abundance variance and infers distribution models sensitive to strain-level variations.
• Strain/sub-strain resolution: Produces abundance estimates with high precision down to sub-strain levels and resolves strain clusters.
• Bias reduction: Addresses bias introduced by similar genome sequences to improve accuracy of abundance estimates.
• Sensitivity and specificity: Enhances detection of minor changes in microbial communities while reducing false-positive rates.
• Differential assessment: Quantifies and performs differential testing of individual taxa within metagenomic samples.

Scientific Applications:

• Strain-level metagenomics: Resolving and quantifying closely related strains within microbial communities from metagenomic sequencing data.
• Differential abundance testing: Detecting subtle abundance changes linked to disease or ecological state by accounting for estimation uncertainty.
• Fine-grained community profiling: Characterizing strain clusters and sub-strain variation for ecological or clinical studies.
• Benchmarking evaluation: Assessing method performance across datasets of varying complexity to compare accuracy and false-positive rates.

Methodology:

DiTASiC applies a generalized linear model to resolve shared read counts and integrates abundance estimation uncertainties by capturing abundance variance and inferring distribution models for differential abundance analysis.