PETAL model documentation
Model families, training data and downloadable files.
Explore the scientific background behind PETAL. Here you find information for each gene regulatory element type we use, the model architecture used for that element, the Plant STARR-seq data behind it and files needed for reproducibility.
Plant STARR-seq, or how we obtain our training data
Plant STARR-seq is a massively parallel reporter assay (MPRA) that enables high-throughput quantification of gene regulatory element activity.
We have extensively applied Plant STARR-seq to characterize
core promoters,
enhancers,
terminators and
silencers.
Combined with advances in deep learning, these rich datasets now provide an opportunity to accurately predict the activity of major gene regulatory elements.
An overview of our models.
The sections below are model-family specific: for each DNA type model we give some architecture notes,
training-data notes and model-performance information.
Note that PETAL contains main and legacy models: main models are the current productive models we suggest to use. Legacy models are old models we keep for reproducibility.
Have a model you would like to make available through PETAL? Please reach out to me via email to discuss whether and how your model could be integrated into or hosted on PETAL.
Enhancer activity prediction
PLACEHOLDER section for enhancer models
Model architecture
PLACEHOLDER
STARR-seq training data
PLACEHOLDER
Citations
Core promoter activity prediction
Placeholder
Model architecture
Placeholder
STARR-seq training data
Placeholder
Citations
Terminator activity prediction
Placeholder
Model architecture
Placeholder
STARR-seq training data
Placeholder