Hi Ian,
I put you here all the bugs that I can encounter using pydock3 dockopt. I suppose that most of them will be non-critical bugs, but it's always interesting for you to have a list of these issues. I will edit this post over time if I find new things.
pydock3 dockopt - init --job_dir_name=JobName command as written in the wiki is not working properly
ERROR: Could not consume arg: --job_dir_name=WDR5_4QL1_dockopt
Usage: pydock3 dockopt - init -
For detailed information on this command, run:
pydock3 dockopt - init - --help
=> Easy fix = ```pydock3 dockopt - init JobName``` works the same way as blastermaster, so I'll edit the wiki accordingly, but I don't know if other "--XXX" arguments work
- Issues with the ```pydock3 dockopt - init - --help``` command - It just runs the dockopt init without printing help, and gives at the end a wrong command :
``` INFO: Showing help with the command 'pydock3 dockopt - init - -- --help'.```
=> Fix : Using ``` pydock3 dockopt ``` command prints help instead
- When resubmitting dockopt run in case of crash (which happends to me with the automatic disconnection of the ssh tunnel), the script scans every 7 to 8 seconds each run already performed, which causes a significant delay
```2022-12-16 08:03:23;INFO;Retrodock job submission skipped for 'dockopt_job_d4f71f41ed21ea891009763ffb61243e_1' since all its OUTDOCK files already exist.
2022-12-16 08:03:30;INFO;Retrodock job submission skipped for 'dockopt_job_d4f71f41ed21ea891009763ffb61243e_2' since all its OUTDOCK files already exist.
2022-12-16 08:03:38;INFO;Retrodock job submission skipped for 'dockopt_job_d4f71f41ed21ea891009763ffb61243e_3' since all its OUTDOCK files already exist.
[...]
2022-12-16 08:10:30;INFO;Retrodock job submission skipped for 'dockopt_job_d4f71f41ed21ea891009763ffb61243e_55' since all its OUTDOCK files already exist.
2022-12-16 08:10:37;INFO;Retrodock job submission skipped for 'dockopt_job_d4f71f41ed21ea891009763ffb61243e_56' since all its OUTDOCK files already exist.
Issue with the name of generated results. Even if you give the program a custom name (in my case "WDR5_4QL1_dockopt"), the results are generated with the default name (in my case dockopt_job_report.pdf)
Very low resolution of the images proposed in the pdf, especially for the first graph. I believe it can be way more interesting to generate svg vector files and merge them in the pdf to not lose picture quality. Maybe it can also be really interesting to generate all these svg files (and/or high-quality png files) in a "results" folder, so people can easily use them in a pptx presentation or make some custom modifications with Adobe Illustrator (even if some png files are already in the "best_retrodock_job" folder).
There is no AUC of the ROC curve, which makes it very difficult or impossible to compare the results with other methods.
In the generated csv file, "dock_executable_path" column does not seem really interesting and adds lots of characters for nothing as it's absolute paths. Can be problematic for people who want to use dockopt intensively and do not want a csv file of several MB.
In the generated csv file, I believe the "," separator can be problematic for some people, as an example if you use it in your paths (I agree that it is stupid, but I know people who do it). A "\t" separator can solve this problem.
It is a bit frustrating for me to not not to have access to the summary of the best poses for each ligand (actives or decoys), with their score and their order of appearance. In my run, the ROC curve seems really bad compared to other docking softwares. As there are no easy generated files allowing me to make my own ROC curve and compare the results to the other methods, I believe that I have to parse OUTDOCK files by myself. If you are interested, I will send you my code too add it in your script toolbox.
Hi Ian, I put you here all the bugs that I can encounter using pydock3 dockopt. I suppose that most of them will be non-critical bugs, but it's always interesting for you to have a list of these issues. I will edit this post over time if I find new things.
pydock3 dockopt - init --job_dir_name=JobName
command as written in the wiki is not working properlyFor detailed information on this command, run: pydock3 dockopt - init - --help
Issue with the name of generated results. Even if you give the program a custom name (in my case "WDR5_4QL1_dockopt"), the results are generated with the default name (in my case dockopt_job_report.pdf)
Very low resolution of the images proposed in the pdf, especially for the first graph. I believe it can be way more interesting to generate svg vector files and merge them in the pdf to not lose picture quality. Maybe it can also be really interesting to generate all these svg files (and/or high-quality png files) in a "results" folder, so people can easily use them in a pptx presentation or make some custom modifications with Adobe Illustrator (even if some png files are already in the "best_retrodock_job" folder).
There is no AUC of the ROC curve, which makes it very difficult or impossible to compare the results with other methods.
In the generated csv file, "dock_executable_path" column does not seem really interesting and adds lots of characters for nothing as it's absolute paths. Can be problematic for people who want to use dockopt intensively and do not want a csv file of several MB.
In the generated csv file, I believe the "," separator can be problematic for some people, as an example if you use it in your paths (I agree that it is stupid, but I know people who do it). A "\t" separator can solve this problem.
It is a bit frustrating for me to not not to have access to the summary of the best poses for each ligand (actives or decoys), with their score and their order of appearance. In my run, the ROC curve seems really bad compared to other docking softwares. As there are no easy generated files allowing me to make my own ROC curve and compare the results to the other methods, I believe that I have to parse OUTDOCK files by myself. If you are interested, I will send you my code too add it in your script toolbox.