A positive score indicates classification as a cancer, a negative score as a non-malignant disease. The column heights are a measure of how well the sample is classified by the SVM algorithm for the 29 genes and the error bars are a measure of the classification variance across the 100 resamplings. SEM: standard error of the mean. A) NSCLC patients: AC-adenocarcinoma, LSCC-lung squamous cell carcinoma, NSCLC-samples not further characterized. B) Non-healthy control samples (NHC) include patients with non-malignant lung disease: COPD: only chronic obstructive pulmonary disease, Benign Nodules: (determined by biopsy), Other: various types of lung diseases C) Receiver-Operator Characteristic curve for classification of samples shown in A and B. AUC: area under the curve. White circle indicates sensitivity-specificity value corresponding to classification score threshold of 0.