Fig. 3: The bio-primed model identifies GAB2 CN as a biologically relevant biomarker for EGFR dependency and drug sensitivity.
From: Bio-primed machine learning to enhance discovery of relevant biomarkers
A Scatter plot shows the correlation coefficient between gene-level CN variation and EGFR dependency (y-axis) across genes sorted by genomic location (x-axis). Points are colored to distinguish adjacent chromosomes. B Scatter plot shows correlation (top), baseline (middle) and bio-primed (bottom) models’ coefficients restricted to chromosome 11. GAB2 is assigned non-zero coefficient in the bio-primed but not in the baseline model. C Boxplots show EGFR dependency (y-axis) across cell lines stratified by combinations of CN gains of EGFR and GAB2. D Barplot shows correlation coefficient between drug sensitivity and GAB2 CN variation (x-axis) for the top 50 negative associations (y-axis). Colors indicate if the drug targets EGFR. E Boxplots shows Afatinib sensitivity (y-axis) across GAB2 amplification status (x-axis) restricted to cell lines with CN neutral EGFR.
