How many heavy atoms can be processed[no reply required]

[webridging]

No reply required

I got it from the paper:

1. excluding glycerine trioleate due to its size (i.e., 65 heavy atoms).
2. 653 neutral compounds with up to 15 heavy (i.e., non-hydrogen) atoms were coarse-grained by the present procedure and subject to TI calculations in both water and octanol environments.
3. To the extent of the system sizes studied here (i.e., up to ∼20 non-hydrogen atoms), this assumption seems to hold, as illustrated in Figure 4. We expect this assumption to break down for larger molecules—the threshold size remains unknown.
4. larger molecules quickly prove challenging. For example, the 22-atom dibenz[a,h]anthracene (Figure 7) required 25 Gb of memory to enumerate all possible mappings (up to the point that extra beads did not further improve the energy function, eq 1) and required ∼1 h of CPU time.
5. The exhaustive enumeration of mappings limits the compound size to roughly 20–25 non-hydrogen atoms on current desktop-computer architectures.

20–25 heavy atoms are recommended.

I found that if there are more than 133 heavy atoms, it can't be processed(That's impossible i know. By the way, i have 64 Gb of memory), here's the configuration I changed：（version changed-vcs, rdkit, python, numpy）