plmDCA

pseudo-likelihood Maximization Direct Coupling Analysis (plmDCA) 2.0

Overview

This package is the GPU-accelerated version of the original code that can be found at PlmDCA.jl. However, this implementation focuses more on the generation rather than contact prediction. It also aims to provide a user-friendly command line interface for training and sampling from a pseudo-likelihood DCA model.

Installation

During the installation, plmDCA will also install adabmDCA and all its dependencies.

To install, clone this repository:

git clone https://github.com/spqb/arDCA.git
cd plmDCA
python -m pip install .

Using the package

We provide a Colab notebook where it is shown hot to train and sample a plmDCA model.

Alternatively, one can install the package locally and run from the command line one of the two implemented routines:

Train plmDCA from the command line

Once installed, you can launch the package routing by using the command plmDCA. All the training options can be listed via

plmDCA train -h

To launch a training with default arguments, use

plmDCA train -d <path_data> -o <output_folder> -l <label>

where path_data is the path to the input multi-sequence alignment in fasta format and label is an identifier for the output files. The parameters of the trained model are saved in the file output_folder/<label>_params.pth, and can be easily loaded afterwards using the Pytorch methods.

By default, the program assumes that the input data are protein sequences. If you want to use RNA sequences, you should use the argument --alphabet rna.

Warning

Depending on the dataset, the default regularization parameters reg_h and reg_J may not work properly. If the training does not converge or the model's generation capabilities are poor, you may want to increase these values.

Note

By default, the training routine returns the non-reciprocal (i.e. asymmetric) version of the model's couplings. To obtain the reciprocal (symmetric) version, add the flag --reciprocal to the script's arguments.

Sample plmDCA from the command line

To generate new sequences using the command line, the minimal input command is

plmDCA sample -p <path_params> -d <path_data> -o <output_folder> -l <label> --ngen <num_sequences>

where num_sequences is the number of sequences to be generated. The output sequences will be saved in fasta format at output_folder/<label>_samples.fasta. Also, a csv log file containing the one and two-site Pearson correlation coefficients between the generated samples and the data as a function of the number of sweeps is returned.

Note

By default, the script will take the first sequence of the input MSA as the wild-type to start the sampling from. To use a random initialization, add the flag --init_random.

License

This package is open-sourced under the Apache License 2.0.

Citation

If you use this package in your research, please cite