LAUSR

Introduction In order to identify new putative oncology drug targets, we have developed a whole-genome CRISPR drop-out screening pipeline. Material and methods The CRISPR drop-out library used in this pipeline is composed of 1 01 094 single guideRNAs targeting ~19 000 genes (5 or 10 guides/gene). To facilitate interpretation of data generated by this pipeline, all cell lines have been extensively characterised by whole-exome sequencing, SNP6 copy number arrays, RNA-sequencing and drug sensitivity testing. A challenge of interpreting CRISPR drop-out data, is the high false-positive rates in detecting essential genes, particularly for those that are within copy number amplified regions of the genome. We have developed a computational tool, CRISPRcleanR, which is capable of identifying and correcting gene-independent responses to CRISPR-Cas9 targeting. Results and discussions We have identified >700 core pan-cancer essential genes using an adaptive computational method. These genes are members of a priori known essential gene sets, as well as being involved in essential biological processes such as cell cycle, DNA synthesis and DNA replication. Methods for associating gene essentiality with genomic features have been developed to understand cellular mechanisms underpinning differential gene essentiality and to identify potential biomarkers for patient stratification. Known dependencies have been identified e.g. PIK3CA is essential in PIK3CA mutant breast cancer cell lines, as well as promising novel associations. Identified associations are incorporated into a weighted prioritisation scoring system integrating clinical, experimental and phenotypic data which is being used to select the most translatable therapeutic targets for further experimental validation. Each potential novel target is also grouped based on tractability potential which further prioritises the most promising novel actionable candidates. Conclusion This project has developed a robust whole-genome CRISPR screening pipeline which has successfully screened a large collection of diverse human cancer cell lines. We have also developed a number of bespoke analytical strategies to process this unique dataset. The integration of multiple evidence types has established a systematic method to extract potential novel actionable targets from data generated by a large-scale screening effort.