Recent News:

• Welcome to ProFactory!

✏️ Table of Contents

• Features
• Supported Models
• Supported Training Approaches
• Supported Datasets
• Supported Metrics
• Reuirement
• Get Started
• Citation
• Acknowledgement

📑 Features

• Vaious protein langugae models: ESM2, ESM-b, ESM-1v, ProtBert, ProtT5, Ankh, etc
• Comprehensive supervised datasets: Localization, Fitness, Solubility, Stability, etc
• Easy and quick data collector: AlphaFold2 Database, RCSB, InterPro, Uniprot, etc
• Experiment moitors: Wandb, Local
• Friendly interface: Gradio UI

🤖 Supported Models

Model	Model size	Template
ESM2	8M/35M/150M/650M/3B/15B	facebook/esm2_t33_650M_UR50D
ESM-1b	650M	facebook/esm1b_t33_650M_UR50S
ESM-1v	650M	facebook/esm1v_t33_650M_UR90S_1
ProtBert-Uniref100	420M	Rostlab/prot_bert_bfd
ProtBert-BFD100	420M	Rostlab/prot_bert_bfd
ProtT5-Uniref50	3B/11B	Rostlab/prot_t5_xl_uniref50
ProtT5-BFD100	3B/11B	Rostlab/prot_t5_xl_bfd
Ankh	450M/1.2B	ElnaggarLab/ankh-base

🔬 Supported Training Approaches

Approach	Full-tuning	Freeze-tuning	LoRA	SES-Adapter
Pre-Training	❎	❎	❎	❎
Supervised Fine-Tuning	❎	✅	✅	✅

📚 Supported Datasets

Pre-training datasets

• CATH_V43_S40 | structures

Supervised fine-tuning datasets (amino acid sequences/ foldseek sequences/ ss8 sequences)

• DeepLocBinary_ESMFold | protein-wise | single_label_classification
• DeepLocBinary_AlphaFold2 | protein-wise | single_label_classification
• DeepLocMulti_ESMFold | protein-wise | single_label_classification
• DeepLocMulti_AlphaFold2 | protein-wise | single_label_classification
• DeepSol_ESMFold | protein-wise | single_label_classification
• DeepSoluE_ESMFold | protein-wise | single_label_classification
• ProtSolM_ESMFold | protein-wise | single_label_classification
• EC_ESMFold | protein-wise | multi_label_classification
• EC_AlphaFold2 | protein-wise | multi_label_classification
• GO_BP_ESMFold | protein-wise | multi_label_classification
• GO_BP_AlphaFold2 | protein-wise | multi_label_classification
• GO_CC_ESMFold | protein-wise | multi_label_classification
• GO_CC_AlphaFold2 | protein-wise | multi_label_classification
• GO_MF_ESMFold | protein-wise | multi_label_classification
• GO_MF_AlphaFold2 | protein-wise | multi_label_classification
• MetalIonBinding_ESMFold | protein-wise | single_label_classification
• MetalIonBinding_AlphaFold2 | protein-wise | single_label_classification
• Thermostability_ESMFold | protein-wise | regression

[!TIP] Only structural sequences are different for the same dataset, for example, DeepLocBinary_ESMFold and DeepLocBinary_AlphaFold2 share the same amino acid sequences, this means if you only want to use the aa_seqs, both are ok!

Supervised fine-tuning datasets (amino acid sequences)

• FLIP_AAV | protein-site | regression
  • FLIP_AAV_one-vs-rest, FLIP_AAV_two-vs-rest, FLIP_AAV_mut-des, FLIP_AAV_des-mut, FLIP_AAV_seven-vs-rest, FLIP_AAV_low-vs-high, FLIP_AAV_sampled
• FLIP_GB1 | protein-site | regression
  • FLIP_GB1_one-vs-rest, FLIP_GB1_two-vs-rest, FLIP_GB1_three-vs-rest, FLIP_GB1_low-vs-high, FLIP_GB1_sampled

📈 Supported Metrics

Metric Name	Full Name	Problem Type
accuracy	Accuracy	single_label_classification/ multi_label_classification
recall	Recall	single_label_classification/ multi_label_classification
precision	Precision	single_label_classification/ multi_label_classification
f1	F1Score	single_label_classification/ multi_label_classification
mcc	MatthewsCorrCoef	single_label_classification/ multi_label_classification
auc	AUROC	single_label_classification/ multi_label_classification
f1_max	F1ScoreMax	multi_label_classification
spearman_corr	SpearmanCorrCoef	regression

✈️ Reuirement

Conda Enviroment

Please make sure you have installed Anaconda3 or Miniconda3.

Hardware

We recommend a 24GB RTX 3090 or better, but it mainly depends on which PLM you choose.

🧬 Get Started

Installation

Quick Start

🙌 Citation

Please cite our work if you have used our code or data.

🎊 Acknowledgement

Thanks the support of Liang's Lab.