MPRA-MNIST Repository

We present MPRA-MNIST: a standardized dataset and toolkit. This resource integrates rigorously preprocessed MPRA data from seminal studies, preserving experimental fidelity while providing:

• Consistent Formats: Ready-to-use sequences, activity scores, and metadata (CSV, FASTA, PyTorch).

• Reproducible Pipelines: Transparent preprocessing code with version-controlled dependencies.

• ML Compatibility: Structured for classification/regression tasks in frameworks like scikit-learn.

By eliminating data-wrangling barriers, MPRA-MNIST enables rapid algorithm validation—shifting focus from technical debt to biological discovery.

Software Requirements

• OS: Ubuntu 20.04.6 LTS x86_64
• CUDA: 12.6
• Python: 3.12.7
• PyTorch: 2.7.1+cu126

Installation

1. Clone the repository:

   git clone https://github.com/autosome-imtf/MPRA-MNIST
   cd MPRA-MNIST

2. Create a Virtual Environment

   conda create -n mpramnist python=3.12.7
   conda activate mpramnist
   pip install torch

3. Install dependencies:

   pip install --upgrade pip
   pip install -r requirements.txt 

4. Install the package in editable mode (for development):

   pip install setuptools wheel
   python setup.py sdist bdist_wheel
   pip install -e .

We have such datasets:

Name	Artcile and link	DOI	Cell types
HUMAN	-----------	-----------	-----------
Agarwal2025	Massively parallel characterization of transcriptional regulatory elements	10.1038/s41586-024-08430-9	HepG2, K562, WTC11
BarbadillaMartinez2026	Regulatory grammar in human promoters uncovered by MPRA-based deep learning	10.1038/s41588-022-01048-5	AGS HAP1 HepG2 K562 MCF7 U2OS HCT116 HEK293 LNCaP
Kircher2019	Saturation mutagenesis of twenty disease-associated regulatory elements at single base-pair resolution	10.1038/s41467-019-11526-w	HepG2, K562, etc
Gosai2024	Machine-guided design of synthetic cell type-specific cis-regulatory elements	10.1101/2023.08.08.552077	HepG2, K562, SK-N-SH
Fromel2025	Design principles of cell-state-specific enhancers in hematopoiesis	10.1038/s41588-022-01048-5	K562, HSPC (7 states)
Sahu2022	Sequence determinants of human gene regulatory elements	10.1038/s41588-021-01009-4	HepG2, GP5D, RPE1
Arensbergen2019	Genome-wide mapping of autonomous promoter activity in human cells	10.1038/nbt.3754	HepG2, K562
Ernst2016	Genome-scale high-resolution mapping of activating and repressive nucleotides in regulatory regions	10.1038/nbt.3678	HepG2, K562
Reddy2023	Strategies for effectively modelling promoter-driven gene expression using transfer learning	10.1101/2023.02.24.529941	JURKAT, K562, THP1
BACTERIA	-----------	-----------	-----------
Evfratov2017	Application of sorting and next generation sequencing to study 5΄-UTR influence on translation efficiency in Escherichia coli	10.1093/nar/gkw1141	The JM109 E. coli strain
Wang2020	Synthetic promoter design in Escherichia coli based on a deep generative network	10.1093/nar/gkaa325	The DH5α E. coli strain
YEAST	-----------	-----------	-----------
Rafi2024	Random Promoter DREAM Challenge Consortium. Evaluation and optimization of sequence-based gene regulatory deep learning models.	10.1101/2023.04.26.538471	strains S288C::ura3, etc
Vaishnav2024	The evolution, evolvability and engineering of gene regulatory DNA	10.1038/s41586-022-04506-6	strains Y8205, S288C::ura3, etc
DROSOPHILA	-----------	-----------	-----------
deAlmeida2022	DeepSTARR predicts enhancer activity from DNA sequence and enables the de novo design of synthetic enhancers	10.1038/s41588-022-01048-5	Drosophila S2

Planned datasets

Priority	Artcile and link	DOI
HUMAN	-----------	-----------
1	Functional dissection of complex trait variants at single-nucleotide resolution	10.1038/s41586-026-10121-6
1	A systematic evaluation of the design and context dependencies of massively parallel reporter assays	10.1038/s41592-020-0965-y
1	Context-dependent regulatory variants in Alzheimer’s disease	10.1101/2025.07.11.659973
1	Massively parallel characterization of regulatory elements in the developing human cortex	10.1126/science.adh0559
1	Multi-scale dissection, compaction and derivatization of mammalian developmental enhancers	10.64898/2026.04.20.719625
2	BRAIN-MAGNET: A functional genomics atlas for interpretation of non-coding variants	10.1016/j.cell.2025.10.029
2	Generative Design of Cell Type-Specific RNA Splicing Elements for Programmable Gene Regulation	10.1101/2025.11.05.686847
2	Fine-tuning sequence-to-expression models on personal genome and transcriptome data	10.1101/2024.09.23.614632
2	Massively parallel characterization of regulatory elements in the developing human cortex	10.1126/science.adh0559
2	Iterative deep learning-design of human enhancers exploits condensed sequence grammar to achieve cell type-specificity	10.1101/2024.06.14.599076
3	Disease-linked regulatory DNA variants and homeostatic transcription factors in epidermis	10.1038/s41467-025-63070-5
3	Deciphering the functional impact of Alzheimer’s Disease-associated variants in resting and proinflammatory immune cells	10.1101/2024.09.13.24313654
3	Uncovering the whole genome silencers of human cells via Ss-STARR-seq	10.1038/s41467-025-55852-8
3	Billion-Scale Deciphering of Human Gene Regulatory Grammar	10.1101/2025.11.10.687627
3	Decoding the MYC locus reveals a druggable ultraconserved RNA element	10.64898/2026.01.29.702547
BACTERIA	-----------	-----------
1	De-novo promoters emerge more readily from random DNA than from genomic DNA	10.1101/2025.08.25.672121
2	Predictive Modeling of Gene Expression and Localization of DNA Binding Site Using Deep Convolutional Neural Networks, this is pre-processed data from Deciphering the regulatory genome of Escherichia coli, one hundred promoters at a time	10.1101/2024.12.17.629042
3	Structure and Evolution of Constitutive Bacterial Promoters	10.1101/2020.05.19.104232
3	The emergence and evolution of gene expression in genome regions replete with regulatory motifs	10.7554/eLife.98654.3
YEAST	-----------	-----------
3	Deep learning of the regulatory grammar of yeast 5′ untranslated regions from 500,000 random sequences	10.1101/gr.224964.117
PLANTS	-----------	-----------
1	Synthetic promoter designs enabled by a comprehensive analysis of plant core promoters	10.1038/s41477-021-00932-y
3	Arabidopsis and maize terminator strength is determined by GC content, polyadenylation motifs and cleavage probability	10.1038/s41467-024-50174-7