Fig. 3From: Capturing the differences between humoral immunity in the normal and tumor environments from repertoire-seq of B-cell receptors using supervised machine learningLetter-value plots are showing distribution of the area under the Receiver Operating Characteristic curve (AUROC) calculated on 89 held-out patient. The figures are illustrating the comparison of (a) different models using amino acid sequences, (b) different trimming/padding strategies, (c) models using various sequence-independent features as well as ensemble model combining them, and (d) CNN against different length of CDR3. Barplots (e) shows the coefficients of linear ensemble modelBack to article page