Fig. 2.

Computational pipeline for data collection and processing. (A) Seven receptors (wild-type Ste2, lower promiscuity HF10, and higher promiscuity Prom6, Prom7, Prom3, Mut1 and TBBI2) were selected. Each of the 55 peptides was converted to vectors using the eight-property VHSE system for a predictor (X) matrix of 104 features. Each receptor was characterized against the peptide library to obtain a dose response curve. Three metrics (EC₅₀, EC₁₀ and response at 10 μM peptide) were obtained from each curve, and further expanded using inverse/logarithm manipulations to obtain the response (Y) matrix. (B) Example grouping schemes for a representative system in which peptides are four residues long and each peptide has three properties (note that our system has 13-residue long peptides with eight properties).