Discovering functional sequences with RELICS, an analysis method for CRISPR screens

Abstract

CRISPR screens are a powerful technology for the identification of genome sequences that affect cellular phenotypes such as gene expression, survival, and proliferation. By targeting non-coding sequences for perturbation, CRISPR screens have the potential to systematically discover novel functional sequences, however, a lack of purpose-built analysis tools limits the effectiveness of this approach. Here we describe RELICS, a Bayesian hierarchical model for the discovery of functional sequences from CRISPR screens. RELICS specifically addresses many of the challenges of non-coding CRISPR screens such as the unknown locations of functional sequences, overdispersion in the observed single guide RNA counts, and the need to combine information across multiple pools in an experiment. RELICS outperforms existing methods with higher precision, higher recall, and finer-resolution predictions on simulated datasets. We apply RELICS to published CRISPR interference and CRISPR activation screens to predict and experimentally validate novel regulatory sequences that are missed by other analysis methods. In summary, RELICS is a powerful new analysis method for CRISPR screens that enables the discovery of functional sequences with unprecedented resolution and accuracy.

Deletion mapping of regulatory elements for GATA3 reveals a distal T helper 2 cell enhancer involved in allergic diseases

“…RELICS can also leverage data from multiple pools to detect FSs that cause smaller changes in gene expression. RELICS has been extensively validated on experimental data and outperforms other tiling CRISPR screen analysis methods 34 . When running RELICS, we used an area of effect model that assumes a spanning deletion efficiency of 25% to account for the variable mutation events generated by NHEJ (Fig.…”

“…We used RELICS (https://github.com/patfiaux/RELICS) 34 to jointly analyze the guide pair counts across all pools and replicates (24 pools total). RELICS leverages a set of known functional sequences, labeled FS0, to estimate sorting parameters, which describe the probability that cells containing different guides will be sorted into each pool.…”

Discovery and Functional Characterization of Pro-growth Enhancers in Human Cancer Cells

“…S2A-B). To identify the critical enhancers driving MYC expression, we used a Generalized Linear Mixed Model (GLMM) framework ( 34 ) to jointly describe the observed pgRNAs counts across sorted pools under two models: a regulatory model (pgRNA targets a regulatory sequence) and a background model (pgRNA does not target a regulatory sequence).…”

“…We selected genomic regions with an averaged CRISPRi score above 5 ad candidate pro-growth enhancers. This threshold was chosen to correspond to FDR < 0.1 based on simulated data from CRSsim ( 34 ). To generate high confidence candidate pro-growth enhancers across different cell types, we further selected the reproducible candidate enhancers, which were identified from two independent pipelines, RELICS v1 and CRISPY.…”

Using CRISPR to understand and manipulate gene regulation

“…CRISPRi tends to induce relatively uniform epigenetic repression of an entire regulatory element, whereas Cas9nuclease can yield variable phenotypes at neighboring sgRNAs because neighboring sgRNAs may fail to disrupt the same TFbinding motif (Hsu et al, 2018;Rajagopal et al, 2016). With these parameters in mind, two pipelines, CRISPR-SURF (Hsu et al, 2018) and RELICS (Fiaux et al, 2020), have been designed to analyze tiled non-coding CRISPR screens. These platforms account for the expected partial correlation among neighboring sgRNAs and allow semi-supervised adjustment of the size of genomic regions that are expected to share phenotypic outcomes.…”