1 min readJan 12, 2018
Response to Marwin Segler (AstraZeneca collaboration)
You wrote:
Looking at figures 5 and 8 our paper http://pubs.acs.org/doi/pdf/10.1021/acscentsci.7b00512 immediately shows that the generated molecules have a similar distribution in chemical space as “real” (ChEMBL) molecules, and are not just clustered in a small area.
It doesn’t immediately show anything. t-SNE visualizations can be misleading:
That’s the kind of things better learnt with online education, I think. Old-fashioned education in the lab has limits.
See also my comment about ETH Zurich.
About open-source and lock-in: see my replies to Vijay and Karl.