eSynth

eSynth is a graph-based molecular synthesis tool written in C++. The author is Dr. Chris Alvin. The source code is shared here with the permission granted. All credit goes to Dr. Chris Alvin, Computational Systems Biology Group, and DeepDrug.ai.

If you find this tool is useful to you, please cite this paper:

Naderi, Misagh, Chris Alvin, Yun Ding, Supratik Mukhopadhyay, and Michal Brylinski. "A graph-based approach to construct target-focused libraries for virtual screening." Journal of Cheminformatics 8, no. 1 (2016): 14.

Updates for esynth2

This version of esynth2 combines the branch updates into a single synthesis workflow that parses fragment inputs, validates naming and extensions, and builds molecules only from accepted fragments. The runtime now supports skipping free-atom inputs, directing output to a chosen directory or file suffix, and constraining synthesis depth with a fragment-level bound.

The builder capabilities were also expanded in this release. esynth2 now includes unique-fragment construction, merged-fragment construction, looser atom-type matching, and optional linker-to-linker connections.

• Removed hard-coded molecular property filters based on Lipinski's Rule of Five, giving users control over which filters to apply for a given study
• Added support for multiple bonds to a single brick atom, eliminating a reconstruction failure mode present in the original
• Added a targeted reconstruction (unique-build) mode for soundness validation, allowing the original molecule to be tested for presence among recombinants without exhaustive enumeration
• Reads connectivity metadata from eMolFrag2's richer SDF output, enabling more general reconstruction including correct handling of super-fragments, free atoms, and unambiguous aromaticity information

Overall, esynth2 reflects the merged branch work as a more configurable and practical tool for large-scale fragment-driven molecule generation.

Command-Line Flags

Use eSynth -usage to print the full built-in help text. The supported flags are:

• -skip-fa skips free-atom input files.
• -usage prints usage information and exits.
• -version prints the version string and exits.
• -unique-build treats all input fragments as unique when constructing molecules.
• -merged-fragments enables unique builds after fragment merging.
• -loose uses looser atom typing when connecting fragments.
• -allow-ll allows linker-to-linker connections.
• -o <file> sets the output file suffix.
• -odir <path> sets the output directory.
• -v <file> enables on-the-fly validation with the given .mol2 file.
• -tc <float> sets the Tanimoto coefficient used for molecule equivalence checks.
• -hl <int> sets the maximum fragment count allowed in a molecule.

Input fragment files are provided as positional arguments.

License

This project is licensed under the MIT License. See the LICENSE file for details.