Methods for designing guide sequences for guided nucleases
Inventors
Sanjana, Neville • Zhang, Feng • Meier, Joshua A.
Assignees
Massachusetts Institute of Technology • New York University • Broad Institute • Harvard University • New York Genome Center
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Abstract
Embodiments disclosed herein provide methods, including computer-implemented methods, for designing guide sequence which may be incorporated into custom, large scale guide sequence libraries. The methods require only a list of target genes as input and utilize on target and off target scores to generate an optimal set of guide sequences for a set of target genes. In certain embodiments, the methods may also utilize multi-tissue RNA-sequencing data and/or protein annotation to design targets to genes that are highly expressed and/or contain a functional protein domain. The invention further comprises guide libraries, cells comprising said guide libraries. Computer-implemented embodiments further improve computer system function by reducing excessive user wait time through the use of data structures that reduce search from linear to logarithmic time.
Core Innovation
The invention describes a computer-implemented method for making guide sequences for a CRISPR-Cas system. Using one or more computing devices, exons in one or more target genes are identified, and for each exon potential target regions are identified that can be targeted with an exon prediction module. Candidate potential target regions are ranked using an off-target avoidance score and an on-target efficiency score determined using a classifier, where the off-target avoidance score is determined at least in part by a cutting frequency determination score.
For each target region above a defined cut-off value, a set of guide sequences is generated that are complementary to the target regions. A set of guide molecules is synthesized that comprise the guide sequences, where each guide molecule is capable of complexing with a CRISPR-Cas system and directing sequence-specific binding of the complex to a corresponding target region. The disclosure relies on exon prediction and classification-based ranking to support generation of a guide sequence set for CRISPR-Cas targeting.
The description additionally includes optional refinements to the guide design workflow, including defining exon or coding-region sets based on consensus coding sequences (CCDS) unions and selecting exons using tissue-specific expression data. It also includes guide-sequence filtering such as removing homopolymers and poly-T termination, and supports generating guide libraries with non-targeting guides as negative controls and synthesis-ready guide constructs.
Claims Coverage
The independent claim covers a computer-implemented workflow for exon-based CRISPR-Cas guide design with 4 main inventive features: exon identification and exon prediction, candidate-region ranking using specified scoring components, guide generation from high-ranking target regions, and synthesis of guide molecules for CRISPR-Cas complexing and binding.
Exon identification with exon prediction module
Identifying, using one or more computing devices, exons in one or more target genes, and for each exon identifying potential target regions that can be targeted using a CRISPR-Cas system with an exon prediction module.
Dual scoring using off-target avoidance and on-target efficiency classifier
Ranking each potential target region based at least in part by an off-target avoidance score and an on-target efficiency score determined using a classifier, wherein the off-target avoidance score is determined at least in part by a cutting frequency determination score.
Generate complementary guides for regions above a defined cut-off
For each target region above a defined cut-off value, generating a set of guide sequences that are complementary to the target regions.
Synthesize guide molecules for CRISPR-Cas complexing and sequence-specific binding
Synthesizing a set of guide molecules comprising the set of guide sequences, wherein each guide molecule is capable of complexing with a CRISPR-Cas system and directing sequence-specific binding of the complex to a corresponding target region.
Defining exons as a union of consensus CDS regions
Defining the exons as a union of consensus coding sequences (CCDS) regions.
Selecting potential target regions based on PAM proximity
Identifying potential CRISPR-Cas target regions based on their proximity to a PAM.
Tissue-specific expression-based selection with normalization by coding region base pair size
Selecting coding regions using tissue-specific expression data by calculating median exon expression per tissue across all samples of the same tissue type and normalizing median read counts by coding region base pair size.
Using a boosted regression tree classifier for on-target efficiency
Implementing the classifier as a boosted regression tree classifier.
Generating non-targeting guides as negative controls
Generating non-targeting guide sequences as a negative control.
Overall, the claims center on exon prediction, classifier-based ranking that combines cutting frequency determination-driven off-target avoidance with on-target efficiency scoring, generation of complementary guide sequences above a cut-off, and synthesis of guide molecules for CRISPR-Cas complexing and sequence-specific binding, with refinements for CCDS-based exon definition, PAM-based candidate identification, tissue-expression selection, a boosted regression tree classifier, and negative-control guides.
Stated Advantages
Reduced user wait time using data structures providing linear-to-logarithmic search time.
Experimental results show increased average targeted exon expression using tissue-expression guidance.
Experimental results show reduced predicted off-target sites using summed CFD-style scores.
Experimental results show increased on-target efficiency after pipeline stages.
Experimental results show improved depletion of essential genes in pooled CRISPR screens.
Experimental results show linear scaling of library generation time with gene count due to precomputation.
Documented Applications
Designing custom, large-scale guide libraries from a target gene list using computer-implemented methods for CRISPR guided-nuclease guide sequence design.
Using tissue-expression guidance such as multi-tissue RNA-seq data to prioritize targeted exons in the guide design pipeline.
Performing pooled CRISPR screens, including depletion of essential genes using pooled guide libraries.
Interested in licensing this patent?