All activity was moved to dev branch!!!
This is a project of Me. Also, let me say in advance that most of files in this repository are stored here just as the first example of our work. It was written using numpy and pytorch and this file doesn't affect our main program.
- Alexey Shmelev - code, algorithms in terms of Programing and Bioinformatics
Here are given three folders: rnn, selam and tasks. Let's define what each of them is dedicated for:
- rnn - this folder store the major part of our work, thus you can find code of the NN (and their saved states) and the important file with settings there
- selam - store the bash script which allow you to quickly generate data to train NN using SELAM generator
- tasks - mainly store .sbatch files which are used to run programs using SLURM
- Create bidirectional RNN for finding part of genome which is under natural selection
- Write working code itself
- Opimize it for better performance in training results
- Create Transformer (regression task) for finding generation where data was taken from, foce of natural selection and percentage of admixture
- Write working code itself
- Opimize it for better performance in training results
- Find optimal mathematical model
- Create Transformer (binary classification) for finding if the natural selection was
- Make program to work parallel
- Create GUI using PyQt
You can compile my programm by your own using Pyinstaller with the following command (file app.spec you can find in this repository)
pyinstaller --noconsole --onefile app.spec
Training results of Transformer (existence of natural selection):
How TOF neural network works:
