Fig. 2

Clustering ORFs based on shared HITs and pairwise similarity of effect on the HT-29 transcriptome. A similarity analysis was applied to the transcriptomics dataset from our expression screen for IBD gene candidates in HT-29 cells. A two-dimensional projection of the similarity in effects between the ORFs was generated using the t-SNE method from the gene-level effects matrix on the log scale, including ORFs with a least 20 upregulated or downregulated HIT genes. ORFs with a similar effect on the transcriptome are clustered together on the graph. Size of the circle for each ORF was made proportional to the number of HITS. Yellow circles represent IBD-causal genes. Arrows were drawn between ORFs with width proportional to a similarity score (see the “Methods” section). Arrows from ORF₁ to ORF₀ represent evidence of shared effects on the subset of the transcriptome defined by the HITS of ORF₀. Only arrows for a positive score greater than a specific threshold (0.6) were drawn. Two clusters can be loosely defined in top right (cluster 1) and bottom left (cluster 2) with little connections between them