search

mhismail / nmga

Genetic algorithm for automated PK/PD model selection
5 stars 1 forks source link

Missing documentation or at least a README #1

Open Janssena opened 2 years ago

Janssena commented 2 years ago

Hey,

Thanks for all the work in providing this application. However, I think it might still be quite difficult to use for people not used to R programming. I'm trying to get the application to work based on the examples in the paper, but am running into some issues.

Some of the things that are not immediately obvious:

  1. The tokens in the alltokens file are evaluated one by one, meaning that $PK {model} $THETA {model}

should be coded as: tokengroup, tokenset, token model, mySet, "thing I want in PK for this Set" model, mySet, "thing I want in THETA for this Set" model, myOtherSet, "thing I want in PK for this Set" model, myOtherSet, "thing I want in THETA for this Set"

  1. I think it is possible to use different number of THETA's for each set by using indexing:

tokengroup, tokenset, token WTonCL, one_param, THETA([1]) WT WTonCL, one_param, (0, 1); param1 WTonCL WTonCL, many params, THETA([1]) WT**THETA([2]) - THETA([3]) WTonCL, many params, (0, 1); param #1 \newline (0, 1); param #2 \newline (0, 1); param #3 \newline

Imagine the \newline actually being newlines. Is this correct synthax? Will the THETA([i]) etc take the ith index from the second token? How does it know that it should take the index from that token? I do not see this behaviour working correctly in the 'Preview' window as of now.

I guess some documentation on how to setup the alltokens.csv in general would be useful.

  1. I am trying to run the program but when clicking 'Genetic Algorithm' in the main window I get the following error:

Error: cannot open the connection. In R I see the following error popping up: Cannot open file GAgens.csv: no such file or directory. This makes sense as that file does not exist. However, it is never generated in the application. I can create it manually (with a Generation, Individual, and Dir column), but then it complains about there not being any lines with information in the file.

(when clicking the View Models button, it similarily complains about allmodsResults.csv.gz)

This raises the question: How are you actually supposed to initialize the program? How do you run the model? I have a ctl file and alltokens.csv, but I guess I need to prepare more files?

Hope you can help with providing this information!

mhismail commented 2 years ago

Hi @Janssena,

Thanks for the detailed report. This shiny app was developed more as a proof of concept and we're currently planning on porting it to another language that doesn't suffer from the same dependency messes and UI limitations as R.

With that said, others have reached out with similar issues as you, and I agree 100% that there should be clear instructions on how to get this app running. I'll work on adding instructions in the README and a complete, functioning demo project.

All the best, Mohamed