APARENT
GitHub
APA Logo

APARENT is a deep neural network that can predict human 3' UTR Alternative Polyadenylation (APA) and the effects of genetic variants on APA. The network was trained on >3.5 million randomized 3' UTR poly-A signals expressed on mini gene reporters in HEK293.

APARENT was described in Bogard et al, Cell 2019 in press.
It was developed in Seelig Lab at the University of Washington.

Contact jlinder2 (at) cs.washington.edu for any questions about the model or data.


Mutagenesis

Mutagenesis
  • 1. Choose a PAS from the genome
  • 2. Predict all possible SNVs
  • 3. Inspect Cleavage alteration
  • 4. Inspect Isoform fold change
Go

APA Isoforms

Mutagenesis
  • 1. Choose APA sites from the genome
  • 2. Specifcy Isoforms
  • 3. Predict Isoform Abundance
  • 4. (Optionally) Alter sequences
Go

PAS Score

Mutagenesis
  • 1. Choose a PAS from the genome
  • 2. Specifcy Isoform
  • 3. Predict PAS Score (strength)
  • 4. (Optionally) Alter sequence
Go

Variants

Variants
  • 1. Choose a PAS from the genome
  • 2. Specify the variant sequence
  • 3. Inspect Cleavage alteration
  • 4. Inspect Isoform fold change
Go

PolyA Detection

Detection
  • 1. Choose pA sites from the genome
  • 2. Or specify a custom sequence
  • 3. Inspect PolyA profile
  • 4. Zoom in and detect PASs
Go

PAS Generation

Generate
  • - 1. Choose generative model
  • - 2. Specify # of sequences to make
  • - 3. Press 'Generate'-button
  • - 4. Inspect generated signals
Go