Domain-PFP

Using Steered Molecular Dynamic Tension for Assessing Quality of Computational Protein Structure Models

Domain-PFP: Protein Function Prediction Using Function-Aware Domain Embedding Representations

Nature Communication Biology, 2023

Nabil Ibtehaz¹, Yuki Kagaya² and Daisuke Kihara^1,2,*

¹ Department of Computer Science, Purdue University, West Lafayette, IN 47907, USA
² Department of Biological Sciences, Purdue University, West Lafayette, IN 47907, USA
^* Correspondence Author: dkihara@purdue.edu

Ibtehaz, N., Kagaya, Y. & Kihara, D. Domain-PFP allows protein function prediction using function-aware domain embedding representations.
Commun Biol 6, 1103 (2023). https://doi.org/10.1038/s42003-023-05476-9

Data

Below are the data used in this work.

File	Description
saved_models.zip	Trained DomainGO-prob model weights and Domain-PFP KNN models
probe_embeddings.zip	Embeddings of the proteins from the PROBE benchmark dataset, computed by DomainGO-prob
cafa3_predictions.zip	GO term prediction by DomainPFP on the CAFA3 dataset
data.zip	Data required to reproduce our experiments. Please refer to Experiments for more information
blast_ppi_database.zip	Database files for function prediction using BLAST and PPI

Source Code

The source codes of Domain-PFP, along with detailed instructions of installation and usage are available in:

https://github.com/kiharalab/Domain-PFP

Online Platform

An easy to use version of Domain-PFP is available to run on Google Colab Notebook.