AMULET

AMULET detects multiplets in single-nucleus ATAC-seq (snATAC-seq) data by analyzing uniquely aligning read counts to distinguish single nuclei from co-isolated nuclei.

Key Features:

• Read count–based methodology: Analyzes uniquely aligning reads across the genome to detect loci with abnormal read counts indicative of multiplets, addressing snATAC-seq data sparsity and limited dynamic range.
• Locus identification: Identifies genomic loci with more than two uniquely aligning reads per nucleus to flag potential multiplet regions.
• Multiplet detection: Compares the number of loci with >2 uniquely aligning reads per nucleus against expected values to classify nuclei as multiplets.
• Implementation and modularity: Implemented as a bash shell script that integrates both steps but allows executing locus identification and multiplet detection independently with adjustable parameters such as q-values.
• Performance metrics: Demonstrates high precision validated by donor-based multiplexing experiments and high recall confirmed via simulated multiplets.
• Optimal read depth requirement: Performs optimally when median valid read depth per nucleus is approximately 25,000 reads.

Scientific Applications:

• snATAC-seq quality control in complex tissues: Identifies and removes multiplets in snATAC-seq datasets from complex or heterogeneous samples such as human blood and pancreatic islets.
• Chromatin accessibility profiling: Reduces multiplet-derived artifacts to improve the reliability of chromatin accessibility analyses.
• Transcription factor and epigenetic mapping: Improves accuracy of transcription factor binding site identification and epigenetic landscape mapping by removing multiplets.

Methodology:

Performs genomic read alignment and statistical analysis by identifying loci with >2 uniquely aligning reads per nucleus and comparing counts of such loci to expected values to detect multiplets.