Automate it by incorporating to STB CLOUD with
in the second quarter of 2023.

Syntekabio currently has its own library of about 20,000 of R-groups. According to the published works in this field, the attempts to attach the R-group to the drug scaffold are carried out independently regardless of the target protein and then brought into the protein’s binding pocket. Contrary to this, Syntekabio has developed methods to enumerate 20,000 R-groups with high activity directly on the mother compound within the pocket. Hence, quantitative structure-activity relationship (QSAR) is no longer necessary as it automatically compares and selects in terms of binding energy. Therefore, Syntekabio's drug optimization is called Auto-Lead-Optimization. It has been operated semi-automatically for now in the local environment, with an aim to fully automate it by incorporating to STB CLOUD within the second quarter of 2023.

Essential Key-challenges for ALGO(Auto Lead-Gen/Opt)

Flow diagram of Lead generation and optimization by DeepMatcher®-Lead. It is showed R-groups are substituted and perturbed at the position of the mother compound, as the top 1,000 candidates are selected through grid cylinder filtering. The process is finalized through MD simulations. (source: cloud.syntekabio.com)

WeakBond Screening by Scaffold Extraction

In this step, the weak binding position is found in the optimal binding pose of the hit compound in the pocket. R-group fragments are used at this position to generate derivatives that bind strongly. While keeping a scaffold from the hit compound in position, we first choose a variety of replaceable regions (red box in the figure) in the hit compound and select one of them which has the lowest interaction efficiency. By removing the regions from the hit compound, we can generate a scaffold. Molecular dynamics (MD) is applied in the same way as shown in hit-discovery. In this step, we select derivatives in the previous step and perform MD simulations.

Grid Cylinder Filtering (GCF)

GCFchecks whether new derivatives have reasonable positions in the pocket without any collisions with theprotein.

RMSDE valuation of 52 Predicted Poses

Finding the optimal conformation after attaching the R group. Results of R group attached to the scaffold in known test-set (RMSD with actual binding pose), showing that most of them are well screened except a few red ones.