RELICS

RELICS identifies functional regulatory sequences from tiling CRISPR screens by applying a Bayesian hierarchical model to analyze sgRNA count data across multiple experimental pools and replicates.

Key Features:

• Bayesian hierarchical modeling: Uses a Bayesian hierarchical model to infer locations and effects of functional sequences from sgRNA counts.
• Overdispersion handling: Accounts for overdispersion in sgRNA count data.
• Overlapping sgRNA effects: Models overlapping effects of multiple sgRNAs that target nearby genomic positions.
• Multi-pool integration: Integrates information across multiple experimental pools and performs joint analysis per replicate.
• Functional sequence estimation: Estimates the number of functional sequences supported by the data.
• CRISPRi/CRISPRa compatibility: Applicable to tiling CRISPR interference (CRISPRi) and CRISPR activation (CRISPRa) screens.
• Performance on simulations: Demonstrated improved precision, recall, and resolution on simulated datasets compared to existing methods.
• Experimental validation: Has been applied to predict and experimentally validate novel regulatory sequences in real datasets.

Scientific Applications:

• Regulatory element discovery: Identification of functional sequences in non-coding genomic regions from tiling CRISPR screens.
• Tiling CRISPR screen analysis: Analysis and interpretation of sgRNA count data from tiling CRISPR experiments.
• Candidate prioritization: Prioritization of candidate regulatory sequences for downstream experimental validation.

Methodology:

RELICS uses a Bayesian hierarchical model to model overlapping sgRNA effects, account for overdispersion in sgRNA counts, integrate information across multiple experimental pools with joint analysis per replicate, and estimate the number of functional sequences supported by the data.