Abstract: The invention provides systems and methods for improved peptide screening library design. In some implementations the systems and methods utilize screening data relating to a plurality of peptides used in a peptide screen against a target molecule to construct a consensus binding sequence alignment using least a subset of the plurality of peptides. For one or more positions of the sequence alignment an observed distance matrix is constructed, the matrix describing a distance between the relative binding activity of pairwise comparisons of each amino acid in a given position. The observed distance matrix is then compared to a plurality of molecular field-based amino acid substitution matrices so as to identify one or more preferred amino acids for use in the design of novel predicted binding peptide sequences for a subsequent peptide screen.

Abstract: Provided are systems and methods for generating context-specific, field based amino acid substitution matrices. In some implementations, the systems and methods utilize a set of characteristics including sequence length, sequence, variable position, backbone conformation, sidechain conformation, and charge and/or ionization state to construct a number of instantiated virtual peptide variants that vary an amino acid at the variable position. Molecular fields are then calculated for each instantiated variant. The fields for each variant are then compared to one another in a pairwise fashion. Values representing the similarity of the fields resulting from the comparison are then assembled into an amino acid substitution matrix.

Abstract: The invention provides systems and methods for generating 3D binding consensus pharmacophores. Initially, peptide screening sequence data is aligned. For one or more positions of the alignment: an observed distance matrix describing a distance between the relative binding activity of pairwise comparisons of each amino acid at the selected position is constructed, the observed distance matrix is compared to a plurality of field-based amino acid substitution matrices having the same shape as the observed distance matrix, preferred amino acid substitution matrices are identified from the plurality of amino acid substitution matrices based on the comparison, and a plurality of characteristics for the selected position are identified using the preferred amino acid substitution matrices. Characteristics for a plurality of positions of the alignment are used to generate three-dimensional peptide structures that represent predicted binding conformations.