Nicolo Fusi | Search bioRxiv

bioRxiv2016

Predicting off-target effects for end-to-end CRISPR guide design

The CRISPR-Cas9 system provides unprecedented genome editing capabilities. However, off-target effects lead to sub-optimal usage and additionally are a bottleneck in development of therapeutic uses. Herein, we introduce the first machine learning-based approach to this problem, yielding a state-of-the-art predictive model for CRISPR-Cas9 off-target effects which outperforms all other guide design services. Our approach, Elevation, consists of two inter-related machine learning models--one for scoring individual guide-target pairs and another which aggregates guide-target scores into a single, overall guide summary score. Through systematic investigation, we demonstrate that Elevation performs substantially better than competing approaches on both of these tasks. Additionally, we are the first to systematically evaluate approaches on the guide summary score problem; we show that the most widely-used method (and one re-implemented by several other servers) performs no better than random at times, whereas Elevation consistently outperformed it, sometimes by an order of magnitude. In our analyses, we also introduce a method to balance errors on truly active guides with those which are truly inactive, encapsulating a range of practical use cases, thereby showing that Elevation is consistently superior across the entire range. We thus contribute a new evaluation metric for benchmarking off-target modeling. Finally, because of the large computational demands of our tasks, we have developed a cloud-based service for end-to-end guide design which incorporates our previously reported on-target model, Azimuth, as well as our new off-target model, Elevation.

Bioinformatics↗

bioRxiv2015

In Silico Predictive Modeling of CRISPR/Cas9 guide efficiency

The CRISPR/Cas9 system provides unprecedented genome editing capabilities; however, several facets of this system are under investigation for further characterization and optimization, including the choice of guide RNA that directs Cas9 to target DNA. In particular, given that one would like to target the protein-coding region of a gene, hundreds of guides satisfy the basic constraints of the CRISPR/Cas9 Protospacer Adjacent Motif sequence (PAM); however, not all of these guides actually generate gene knockouts with equal efficiency. Leveraging a broad set of experimental measurements of guide knockout efficiency, we introduce a state-of-the art in silico modeling approach to identify guides that will lead to more effective gene knockout. We first investigated which guide and gene features are critical for prediction (e.g., single- and di-nucleotide identity of the gene target), which are helpful (e.g., thermodynamics), and which are predictive but redundant (e.g., microhomology). We also investigated evaluation measures for comparing predictive models in the present context, suggesting that Area Under the Receiver Operating Curve is not ideal. Finally, we explored a variety of different model classes and found that use of gradient-boosted regression trees produced the best predictive performance. Pointers to our open-source software, code, and prediction server will be available at http://research.microsoft.com/en-us/projects/azimuth.

Bioinformatics↗