dataset_info:
features:
- name: seq
dtype: string
- name: 'y'
dtype: float64
splits:
- name: train
num_bytes: 704490
num_examples: 1021
- name: val
num_bytes: 151110
num_examples: 219
- name: test
num_bytes: 151110
num_examples: 219
download_size: 471183
dataset_size: 1006710
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: val
path: data/val-*
- split: test
path: data/test-*
license: mit
📊 mRFP expression dataset
This dataset comprises E. coli protein expression data for monomeric Red Fluorescent Protein (mRFP). It was generated from a synonymous codon-randomized library with 1,459 gene variants in E. coli, with a measured protein expression level for each. The original dataset is from CodonBERT.
⁉️ Dataset Contents
- Sequence: The mRNA sequence of the mRFP variants.
- Expression: The measured protein expression level.
🎯 Purpose
This dataset serves as a benchmark for fine-tuning models on a regression task, predicting expression behavior from its sequence. We used this dataset to fine-tune CDS-BART, a BART-based foundation model trained on massive mRNA sequences. Demonstrating its ability to perform downstream tasks related to mRNA regulation which are fine-tuned for various mRNA-related downstream task. CDS-BART available at GitHub
🔧Usage
from datasets import load_dataset
dataset = load_dataset('mogam-ai/CDS-BART-mRFP-expression')
📚 Dataset Reference
- Nieuwkoop, Thijs, et al. "Revealing determinants of translation efficiency via whole-gene codon randomization and machine learning." Nucleic acids research 51.5 (2023): 2363-2376.
- Li, Sizhen, et al. "CodonBERT large language model for mRNA vaccines." Genome research 34.7 (2024): 1027-1035.