Closed tagirshin closed 9 months ago
I believe this is an old bug that has to do with the incorrect comparison operator being used. In some places in the code it should be "<" but it is "<=" that is implemented. We might update this, but in principle it doesn't affect the algorithm. One just has to reimagine what "max_transforms" is.
What is the "number_of_steps" you are referring to?
For example, when I read hdf5 file with results and statistics of a tree search, I have these columns:
So here I don't understand why max_transforms and number_of_steps are different
P.S. sorry for the long answer
max_transforms
refers to the depth of the MCTS search tree, i.e. it is a function over all routes extracted.
number_of_steps
refers to the depth of top-ranked route
I will create an issue to add proper documentation for the output dictionaries.
Fixed by #141
Hello! During analysis of tree search results I found out that if I use standard config without specifying max_transforms parameter, the tool can give me routes with 7 transformations instead of standard 6. Here is an example of a found pathway:![route000](https://user-images.githubusercontent.com/17476642/195206552-84a2be8c-aca8-42c0-8d70-a45b54e124fe.png)
I tried to read the config file and on this side everything is fine:
After I tried to browse through the code, but didn't catch the problem right away. I use the last version of aizynthfinder, taken from github.
Also, regarding this issue, I don't understand how max_transforms and number_of_steps are related. From the docs I've got, that max_transforms is the max depth of the MCTS tree, but why then number_of_steps is needed?