The invention relates to a molecular
virtual screening method based on
protein sequence alignment, which comprises the following steps: according to an input
protein sequence to be subjected to
molecular screening, acquiring multi-sequence
coupling information of
protein by using an HHbits program; the frequency PSFM of the same residues appearing at the corresponding positions of the protein sequence to be predicted and the multi-sequence
coupling information being calculated; by using the same method, generating the PSFM of each protein sequence in a protein-ligand interaction
database BioLiP; calculating the residue alignment
score and
similarity matching quality of the to-be-predicted protein and each protein in BioLiP, and obtaining a potential seed molecule set according to the matching
quality score; and calculating the sum of the two-dimensional
fingerprint values of each molecule in the molecule
database and all the molecules in the seed molecule set, sorting all the molecules in the
DrugBank according to scores, and obtaining x.NDrugBank molecules with the top scores as a
molecular sieve selection set of protein sequences to be subjected to
molecular screening. The method can be used for any screening scene.