Fig. 2

CRISPRi screen identifies GC essential transcript isoforms. a Transcripts with gain or loss (compared to adjacent normal) of H3K4me3 signal in GC. Fold change is calculated by comparing H3K4me3 signals in tumour and adjacent normal tissue. H3K4me3 signals are an average of 17 tumour or normal signals from [10]. b Example H3K4me3 profiles of a gained GC transcript, TRPM2. c GSEA showing the top-scoring enriched pathways of transcripts with gain or loss of H3K4me3. d Scheme describing the isoform-specific CRISPRi screen. A library containing 8852 sgRNAs  targeting 820 GC gained transcripts was used to identify transcripts that are essential for the proliferation of normal gastric (HFE145) or 5 GC cell lines. e Distribution of sgRNAs targeting negative controls or core essential genes. f Expression levels of AXIN2, a WNT target gene, in GC cell lines. g Violin plot showing CTNNB1 dependency in GC cell lines. Dots represent individual CTNNB1 targeting sgRNAs. pValue between WNT-active and WNT-inactive cell lines was calculated using two-tailed unpaired t test. (*p ≤ 0.05). h–m MAGeCK analysis [22] was used to identify transcript isoforms that are essential in a normal gastric cell line (HFE145) or 5 GC cell lines. GC gained transcripts are in blue, and unaltered transcripts are in grey. Pan-essential genes (genes that score as essential in ≥ 90% of cell lines in DepMap) are indicated with a red border