This repository contains the code for the training, inference, and interpretation of Massively Parallel Reporter Assay (MPRA) prediction models.
This project utilizes a transfer-learning approach. Our source models (originally trained to predict chromatin tracks or MPRA activity in other cell types from DNA sequences) are adapted from the following works:
- Sei: Chen, K.M., Wong, A.K., Troyanskaya, O.G. et al. A sequence-based global map of regulatory activity for deciphering human genetics. Nat Genet 54, 940–949 (2022). https://doi.org/10.1038/s41588-022-01102-2
- Malinois: Gosai, S.J., Castro, R.I., Fuentes, N. et al. Machine-guided design of cell-type-targeting cis-regulatory elements. Nature 634, 1211–1220 (2024). https://doi.org/10.1038/s41586-024-08070-z
To set up the environment and install the package locally:
git clone [https://github.com/osyafinkelberg/mpra-predictor.git](https://github.com/osyafinkelberg/mpra-predictor.git)
cd mpra-predictor
python3 -m venv .venv
source .venv/bin/activate
pip install -e .The example code in this repository was trained on MPRA data covering ~60,000 200bp tiles derived from viral genomes. The assays were performed in 6 cell lines (GM12878, Jurkat, MRC5, A549, HEK293, K562).
Data Source: Tommy H. Taslim, Joseph A. Finkelberg, et al. (2025). Global cis-regulatory landscape of double-stranded DNA viruses. bioRxiv: 10.1101/2025.07.20.665756v1