pi-PrimeNovo-PTM

PTM Model

This repository contains the code and weights for a PrimeNovo model finetuned on "Phosphorylation (+79.97)" data. The model was adapted from the original PrimeNovo (trained on MassiveKB) by finetuning it on the 2020-Cell-LUAD dataset. The training data was enriched with phosphorylation data, so the model is designed for detecting phosphorylation PTMs.

Environment Setup

Set up the environment exactly as before to ensure compatibility.

Model Overview

The finetuned model has been modified to predict one additional token compared to the original PrimeNovo. For convenience, we added an extra PTM token "B" in the config.yaml located in this directory. When the model outputs a "B", it represents "Phosphorylation (+79.97)".

You can also train or finetune your own PrimeNovo-PTM model by reassigning the token "B" to a different PTM of your choice. Important: If you change the token "B" to another value, ensure that your training data reflects this change so that the target PTM matches the modified token during training.

Modifying the PTM Token

If you decide to change the PTM from "Phosphorylation (+79.97)" to a different modification, update the PTM molecular weight in the following files:

1. in mass_con.py: change variable mass_b =79.9663 to whatever new PTM mole weight it has.

2. in model.py: change variable mass_b =79.9663 to whatever new PTM mole weight it has.

3. in transformers.py: change variable mass_b =79.9663 to whatever new PTM mole weight it has.

4. in config.yaml: change "B": 79.9663 to whatever new PTM mole weight it has.

And that's it, you can then finetune on your own dataset with additional 1 PTM, just replace that PTM with letter "B" anywhere it appears in your training/testing data.

Run Instructions

Note!!!!!!!!!!!!!!!!!!: All the following steps should be performed under the main directory: pi-PrimeNovo-PTM. Do not use cd .. !!!!!!!!!!!!!!!!!!!

Step 1: Download Required Files

To evaluate the mgf containing Phosphated PTM:

1. Model Checkpoint: PTM_Phosphorylation.ckpt

Note: If you are using a remote server, you can use the gdown package to easily download the content from Google Drive to your server disk.

Step 2: Choose the Mode

The --mode argument can be set to either:

• eval: Use this mode when evaluating data with a labeled dataset.
• denovo: Use this mode for de novo analysis on unlabeled data.

Important: Select eval only if your data is labeled.

Step 3: Run the Commands

Execute the following command in the terminal:

python -m PrimeNovo.PrimeNovo --mode=eval --peak_path=./PTM_test.mgf --model=./PTM_Phosphorylation.ckpt

This automatically uses all GPUs available in the current machine.

Step 4: analyze the output

We include a sample running output ./output.txt. The performance for evaluation will be reported at the end of the output file.

If you are using denovo mode, you will get a denovo.tsv file under the current directory. The file has the following structure:

label	prediction	charge	score
Title in MGF document	Sequence in ProForma notation	Charge, as a number	Confidence score as number in range 0 and 1 using scientific notation

The example below contains two peptides predicted based on some given spectrum:

label	prediction	charge	score
MS_19321_2024_02_DDA	ATTALP	2	0.99
MS_19326_2024_02_DDA	TAM[+15.995]TR	2	0.87

NOTE: "B" amino acid will stands for the additional PTM added to this code, in this case it is Phosphorylation.

Citation

Bibtex：

@article{zhang2025pi,
  title={$\pi$-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing},
  author={Zhang, Xiang and Ling, Tianze and Jin, Zhi and Xu, Sheng and Gao, Zhiqiang and Sun, Boyan and Qiu, Zijie and Wei, Jiaqi and Dong, Nanqing and Wang, Guangshuai and others},
  journal={Nature Communications},
  volume={16},
  number={1},
  pages={267},
  year={2025},
  publisher={Nature Publishing Group UK London}
}

Standard Citation:

Zhang, X., Ling, T., Jin, Z. et al. π-PrimeNovo: an accurate and efficient non-autoregressive deep learning model for de novo peptide sequencing. Nat Commun 16, 267 (2025). https://doi.org/10.1038/s41467-024-55021-3