Whole-genome GFP-TFEB localization screen. (a) GSEA pathway analysis annotated using the gene sets derived from the GO Cellular Component database of the Molecular Signatures Database. On the x-axis is the GSEA normalized enrichment score, and the color of the bars represents the GSEA calculated FDR probabilities. (b) Volcano sgRNA plot comparing sgRNA abundance in the photoactivated samples following 8 h of starvation to sgRNA abundance in the unsorted sample. Vertical red line is set at the log₂ fold-change threshold; horizontal red line is set at the Benjamini-Hochberg corrected P value of 15%. Genes selected for secondary screening are shown in green. See also Table S2. (c) Top candidates from the primary screen were selected for secondary screening. The GFP-TFEB localization CNN model probability value was used for measuring the perturbation effect on GFP-TFEB localization over time during starvation. Heatmap including all genes included in secondary screen; low probability values are shown in purple for cytosolic TFEB-GFP, high probability values (yellow) for nuclear TFEB-GFP. n = 3, *P < 0.05, **P < 0.01, or ***P < 0.0001 obtained using repeated-measures ANOVA test. P value, one sgRNA was significant. Data distribution was assumed to be normal, but this was not formally tested. (d) Triplicates of 350 images per sgRNA knockdown were imaged for 18 h during starvation. The AI-PS segmentation and CNN prediction was deployed to compute the translocation value (y-axis). GFP-TFEB translocation dynamics observed during starvation for selected gene candidates, nontargeted sgRNA in black, gRNA targeting the designated protein in red. Quantification is displayed as mean ± SEM from three independent experiments.