Coded by Galymzhan Moldagulov (Grad student @ IBS-CARS & UNIST, KR)
Big shoutout to Ph.D. Yaroslav Sobolev (IBS-CARS, KR) for teaching me how embeddings of multi-dimensional space work, Junren Li (PKU, CN) for his guidance with Faerun, as well as Sanzhar Nurgaliyev (NU, KZ), and Assanali Salem (UNIST, KR) for their help!
You can easily:
- explore chemical space in the interactive Faerun map;
- search complexes by their CSD code;
- switch between color schemes in the plot legend;
- look at 3D structures of individual complexes (e.g. WAXGAG, KIPBAP and AGUZAD).
- Sampled 10-108k mononuclear organometallic complexes from tmQM dataset and CSD [1, 2];
- Vectorized molecules were embedded/projected into 2D TMAP representation [3];
- Molecules were vectorized by MHFP (MinHash Fingerprint)[4];
- Chemical space was visualized with interactive Faerun module [5];
- Visualization of individual molecules in various 3D representations was impleted using 3Dmol.js [6].
- J. Chem. Inf. Model. 2020, 60, 12, 6135–6146;
- Acta Cryst. 2016, B72, 171-179;
- J. Cheminform., 2018, 12, 12;
- J. Cheminform., 2018, 10, 66;
- Bioinformatics, 2018, 34, 8, 1433–1435;
- Bioinformatics, 2015, 31, 8, 1322–1324.