5. Small-molecule Design and Prediction Tools

BioMoDes: A Repository of Tools for Biomolecular Modeling and Design

This page of BioMoDes lists state-of-the-art and emerging tools for Small-molecule Design and Prediction.

5.1. Ligand Design

2024 (Click to collapse/expand)
  • POLYGON: A model for generating novel compounds that inhibit multiple proteins. POLYGON uses a neural network with a reinforcement learning component for the generation of inhibitors targeting two different proteins of interest (i.e. polypharmacology inhibitors/drugs).
    Published: March 18, 2024
    Paper | Code (GitHub)

  • PMDM: A diffusion model for generating 3D ligands given a target protein pocket.
    Published: March 26, 2024
    Paper | Code (GitHub)

  • SyntheMol: A generative AI model for designing novel synthesizable compounds with antibacterial activity.
    Published: March 22, 2024
    Paper | Code (GitHub)

  • CLigOpt: A fragment-based drug design model to generate ligands with desired properties from a given fragment pair.
    Posted: March 17, 2024

  • PocketFlow: A structure-based model for generating novel ligands given a protein binding pocket. PocketFlow shows SOTA performance and generates 100 % chemically valid, drug-like compounds, ligand binding pose close to that predicted by molecular docking, and active hits against HAT1 & YTHDC1 validated with wet experiments
    Published: March 11, 2024
    Paper | Code (GitHub)

5.2. Ligand and Linker Design

  • DiffLinker: An equivariant diffusion model for linker design between molecular fragments to obtain chemically relevant compounds. Given a set of arbitrary number of disconnected fragments, DiffLinker places suitable atoms in-between to generate molecules incorporating all the initial fragments.
    Published: April 11, 2024
    Paper | Demo (HuggingFace) | Code (GitHub)

5.3. Ligand Property Prediction

  • ProSmith: A multimodal network and webserver for predicting drug-target interaction/affinity, small-molecule enzyme substrate, and enzyme Km for target substrate.
    Posted: Oct 12, 2023
    Preprint | Code (GitHub) | Webserver

5.4. Ligand Conformer Generation

  • PIDM: A physics-informed diffusion-based model for small-molecule conformer generation.
    Published: March 14, 2024
    Paper | Code (GitHub)

I try my best to make the information on this website as accurate as possible. If you find any errors in the contents of this page or any other page on this website, I would greatly appreciate that you kindly get in touch with me at

If you are interested in joining my free weekly “BioMoDes and Top Reads” newsletter, please subscribe below.

* indicates required