Integrating synthetic accessibility with AI-based generative drug design

Abstract

Generative models are frequently used for de novo design in drug discovery projects to propose new molecules. However, the question of whether or not the generated molecules can be synthesized is not systematically taken into account during generation, even though being able to synthesize the generated molecules is a fundamental requirement for such methods to be useful in practice. Methods have been developed to estimate molecule "synthesizability", but, so far, there is no consensus on whether or not a molecule is synthesizable. In this paper we introduce the Retro-Score (RScore), which computes a synthetic accessibility score of molecules by performing a full retrosynthetic analysis through our data-driven synthetic planning software Spaya, and its dedicated API: Spaya-API (https://spaya.ai). We start by comparing several synthetic accessibility scores to a binary "chemist score" as estimated by chemists on a bench of generated molecules, as a first experimental validation that the RScore is a reliable synthetic accessibility score. We then describe a pipeline to generate molecules that validate a list of targets while still being easy to synthesize. We further this idea by performing experiments comparing molecular generator outputs across a range of constraints and conditions. We show that the RScore can be learned by a Neural Network, which leads to a new score: RSPred. We demonstrate that using the RScore or RSPred as a constraint during molecular generation enables our molecular generators to produce more synthesizable solutions, with higher diversity. The open-source Python code containing all the scores and the experiments can be found on ( https://github.com/iktos/generation-under-synthetic-constraint ).

Keywords: In silico molecular generation; In-silico synthesizability; Retrosynthesis artificial intelligence; machine learning.

Fig. 1

Imposed structure for Pi3K/mTOR generation

Fig. 2

Normalized histogram of the RScore1min on molecules from Chembl dataset

Fig. 3

Correlation between the RScore1min and the number of synthetic steps given by Spaya API on a sample from Chembl dataset

Fig. 4

Histogram of RA score on the Chembl dataset

Fig. 5

Correlation between RA score and RScore1min on Chembl dataset

Fig. 6

Correlation between SA score and RScore1min on Chembl dataset

Fig. 7

Correlation between SC score and RScore1min on Chembl dataset

Fig. 8

Correlation between the RScore1min and the values predicted from the neural network on a test set

Fig. 9

Reward and accessibility of the top 100 molecules for each task and with different synthetic constraints. The red line is the average reward (without the synthetic score) on the top 100 molecules of the generation. The green line is the percentage of the top 100 molecules with a RScore3min above or equal to 0.5

Fig. 10

Number of molecules in the blueprint for each generation, with indication on their RScore3min range

Fig. 11

Example of a synthesis route obtained by Spaya