Predicting protein functions using positive-unlabeled ranking with ontology-based priors

Abstract

Automated protein function prediction is a crucial and widely studied problem in bioinformatics. Computationally, protein function is a multilabel classification problem where only positive samples are defined and there is a large number of unlabeled annotations. Most existing methods rely on the assumption that the unlabeled set of protein function annotations are negatives, inducing the \emph{false negative} issue, where potential positive samples are trained as negatives. We introduce a novel approach named PU-GO, wherein we address function prediction as a positive-unlabeled classification problem. We apply empirical risk minimization, i.e., we minimize the classification risk of a classifier where class priors are obtained from the Gene Ontology hierarchical structure. We show that our approach is more robust than other state-of-the-art methods on similarity-based and time-based benchmark datasets

Dependencies

• Python 3.10
• PyTorch 2.1.0
• FAIR-ESM (for predicting over FASTA files)
• Other basic dependencies can be installed by running: conda env create -f environment.yml
• Install diamond program on your system (diamond command should be available)

Data Availability

• Download our data from https://dx.doi.org/10.5281/zenodo.11079885 - Here you can find the data used to train and evaluate our method.

• You will get a file named pu-go-data.tar.gz. Place it under the directory containing this repository

• Uncompress the data with tar -xzvf pu-go-data.tar.gz

• For each subontology directory [mf, cc, bp] run the following command to extract the 10 trained models:

cd data/mf
tar -xzvf models.tar.gz

Scripts

• pu_go.py script to train/test PU-GO on the similarity-based split. We show the commands with the selected hyperparameters for each subontology.

  • MFO: python pu_go.py -dr data/ -ont mf --run 0 --batch_size 35 --loss_type pu_ranking_multi --margin_factor 0.01152 --max_lr 0.004573 --min_lr_factor 0.0008792 --prior 0.000143 -ld
  • CCO: python pu_go.py -dr data/ -ont cc --run 0 --batch_size 30 --loss_type pu_ranking_multi --margin_factor 0.09812 --max_lr 0.0008681 --min_lr_factor 0.08364 --prior 0.0001106 -ld
  • BPO: python pu_go.py -dr data/ -ont bp --run 0 --batch_size 39 --loss_type pu_ranking_multi --margin_factor 0.02457 --max_lr 0.0004305 --min_lr_factor 0.09056 --prior 0.0007845 -ld

• pu_go_time.py script to train PU-GO on the time-based split.

• predict.py script to predict functions given a FASTA file.

  • Usage: python predict.py -if your_fasta_file.fa -d cuda

Note: We used the ESM2-15B model using a NVIDIA A100 GPU.

Citation (TODO)