Skip to main content
Fig. 2 | BMC Genomics

Fig. 2

From: A machine learning classifier trained on cancer transcriptomes detects NF1 inactivation signal in glioblastoma

Fig. 2

Performance of our classifier on a validation set. a Two distinct western blots for each of our twelve samples. The controls are U87-MG, an NF1 WT glioblastoma cell line that exhibits proteasomal degradation of the NF1 protein. U87 + PI are U87-MG cells are treated with the proteasome inhibitors (PI) MG-132 and bortezomib to block proteasome-mediated degradation of NF1. We used the NF1/tubulin ratio normalized to U87 + PI as our NF1 protein level estimate. b Prediction scores for each of the 500 classifiers weighted by cross validation test set AUROC where a negative number indicates NF1 wildtype and a positive number is indicates NF1 inactivation. Darker shades of blue indicate higher observed NF1 protein concentrations. c We quantify protein against U87 + PI and provide the mean of the weighted predictions. d Based on weighted predictions, we show the abundance of NF1 protein compared to U87 + PI

Back to article page