Fig. 4

Flowchart to illustrate MUMAL2’s framework. Ten different values for the cost of a false positive are tested in the cost matrix. After selecting the best value in terms of the resultant sensitivity, the final model is built, including the use of TSA with probability range correction. The probability threshold for a 1% FDR identified in the ROC analysis is converted to 0.5 by the TSA. Therefore, the end model uses 0.5 as the discriminant probability, i.e., the set of PSMs with FDR = 0.01 are characterized as the ones with high probabilities (≥ 0.5)