Add enrichment plots of local minima to DockOpt job report

[ianscottknight]

Seth Vigneron (seth.vigneron@ucsf.edu) suggests the following:

"Another suggestion for dockopt might be to also include the enrichment plots for local maxima in the dockopt_job_report.pdf. Below is the first the image of the best job, and then the also for the one at (0.9,1.2) which had a lower enrichment score but actually better early enrichment. I think people would appreciate being able to easily see, or know how to easily see the curves for any given dockopt job."

[jir322]

great suggestion. This speaks to "semi-automatic" mode. When there is full optimization, ideally, the system would automatically figure out which is better. The fact that better early enrichment is not recognized as such by the metric used for optimization, suggests to me that we have the metric wrong. two ideas (which probably do not represent a consensus view in the lab!) 1. compute enrichment score for the first 50% of the graph, favoring early enrichment. could be called "early enrichment score" 2. some other arbitrary function that upweights the early enrichment. thus e.g. 4 x first quartile, 2 x second quartile, 1 x third quartile, 0.5 x 4th quartile. Quartiles moving horizontally from left to right, obviously. We would like to arrive at a fully automatic solution, but I understand the wish to support semi-automatic approaches. John Irwin UCSF Pharmaceutical Chemistry http://irwinlab.compbio.ucsf.edu

…

On Thu, Dec 1, 2022 at 8:55 PM Ian Scott Knight ***@***.***> wrote: Seth Vigneron ***@***.***) suggests the following: "Another suggestion for dockopt might be to also include the enrichment plots for local maxima in the dockopt_job_report.pdf. Below is the first the image of the best job, and then the also for the one at (0.9,1.2) which had a lower enrichment score but actually better early enrichment. I think people would appreciate being able to easily see, or know how to easily see the curves for any given dockopt job." <https://user-images.githubusercontent.com/25094252/205217051-c5221056-17f8-445d-afe0-b03b3aab0ff3.png> <https://user-images.githubusercontent.com/25094252/205217066-469e2058-8ccb-4425-87b4-6db7f3b3f37f.png> <https://user-images.githubusercontent.com/25094252/205217081-eff998fd-c543-487b-b91f-a9fc732d585d.png> — Reply to this email directly, view it on GitHub <#22>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABIR2H4D4JIEP7FP23WLZRDWLF6LDANCNFSM6AAAAAASRQLO2M> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>