Automatic conformation analysis method for quasi-drug organic molecules

An automatic conformation analysis method for quasi-drug organic molecules, including the following steps:

• (1) extracting a group of fragments from an input molecule, wherein there are primarily three types of fragments:
• a flexible bond fragment, a ring isomerism fragment, and a conformation isomerism fragment, wherein:
• the flexible bond fragment includes a flexible bond or two coupled flexible bonds, and an adjacent chemical group beside the flexible bonds, and whether or not two flexible bonds are in a coupled relation is determined based on a topological rule;
• a potential energy surface or extreme points on the flexible bond fragment are searched out by further analysis on the flexible bond fragment, so as to represent a potential energy surface of the whole molecule on this flexible bond;
• the ring isomerism fragment includes a non-conjugate ring or a condensed ring consisting of a plurality of rings;
• the conformation isomerism fragment includes one or more cis-trans sites or chiral centers, and a surrounding chemical environment;
• (2) carrying out knowledge-based conformation recommendation, molecular conformations including the corresponding fragments extracted in step (1) are searched out from a pre-established conformation database, and then conformation statistics of a specific flexible bond or ring isomerism, and a specified cis-trans or chiral center selected from the molecule are carried out to obtain recommended conformations and confidence coefficients of the recommended conformations;
• the confidence coefficients comprise an empirical threshold, if the confidence coefficients are greater than the empirical threshold, a credibility of the knowledge-based recommendation is considered as high, and the recommended conformations are directly used;
• or if the confidence coefficients are small, the conformation analysis continues to be carried out in a more accurate way;
• (3) carrying out conformation recommendation based on force field scanning, wherein scanning strategies comprise:
• one-dimensional or two-dimensional equal-interval rotational scanning of flexible bonds, conformation scanning of flexible rings, and cis-trans or optical configuration scanning;
• based on a molecular force field, conformations generated through the scanning strategies are rapidly optimized, and energy of the conformations is calculated;
• then reasonable conformations are screened out according to the energy;
• (4) verifying, by quantum mechanics (QM), the conformations generated in step (3);
• if the verification succeeds, it indicates that the conformations recommended in step (3) are valid;
• or otherwise, the force field needs to be corrected next;
• (5) correcting the force field:
• particularly, if a universal force field fails to accurately describe the chemical environment of the molecule, force field parameters should be improved to adapt to the molecule;
• upon analysis on the corresponding fragments, QM sampling computation is additionally conducted according to the flexible bond, the ring isomerism and the conformation isomerism to serve as a training set for correcting the force field parameters, and then the force field parameters are corrected;
• the force field is scanned again after the force field parameters are corrected;
• (6) collecting conformation lists of the fragments, the conformation lists separately recommended by the fragments are collected;
• values expressed by the conformations (including dihedral values of the flexible bonds or isomerous space parameters of the ring isomerism, and cis-trans or chiral identifications) are extracted to be used for subsequent parameter combination and optimization;
• and
• (7) combining and optimizing conformation parameters of the fragments through a genetic algorithm, and finding out a set of optimal conformations through global searching.