Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Proteomic analysis reveals distinct gene regulatory functions of the paralogs MAGOH and MAGOHB in cell proliferation.

Biochimica et biophysica acta. Gene regulatory mechanisms·2026
Same author

CDK12/CDK13 inhibition disrupts transcriptional elongation and replication fork progression in glioblastoma.

EMBO molecular medicine·2026
Same author

An EJC-independent novel isoform of MAGOH: the MAGOH-Δ37 isoform and its interactome capture.

Biochemical and biophysical research communications·2025
Same author

BioID proximity mapping reveals novel SAP18 interactions in the prespliceosomal complex.

Biochemical and biophysical research communications·2024
Same author

Evaluation of the Effect of Parental Smoking on Gingival Melanin Pigmentation in Children.

International journal of clinical pediatric dentistry·2024
Same author

A Comparative Evaluation of Nanosilver Fluoride, Chlorhexidine, and Sodium Fluoride When Used as a Varnish on <i>Streptococcus mutans</i> Levels in Children with Caries.

International journal of clinical pediatric dentistry·2024

Related Experiment Video

Updated: Jan 2, 2026

Bidirectional Retroviral Integration Site PCR Methodology and Quantitative Data Analysis Workflow
12:53

Bidirectional Retroviral Integration Site PCR Methodology and Quantitative Data Analysis Workflow

Published on: June 14, 2017

11.1K

Using the Chou's 5-steps rule to predict splice junctions with interpretable bidirectional long short-term memory

Aparajita Dutta1, Aman Dalmia2, Athul R2

  • 1Department of CSE, Indian Institute of Technology, Guwahati, India.

Computers in Biology and Medicine
|November 30, 2019
PubMed
Summary

This study evaluates visualization techniques for interpreting recurrent neural network (RNN) models in genome sequence analysis. Perturbation-based methods excel at identifying canonical splice motifs, while back-propagation methods are better for non-canonical ones.

Keywords:
AttentionBidirectional long short-term memory networksIntegrated gradientsOcclusionOmissionSmooth gradientsSplice junction predictionVisualization

More Related Videos

Using Human Differentially Expressed Gene Lists to Perform Downstream Pathway Enrichment Analysis and Target Prioritization
03:08

Using Human Differentially Expressed Gene Lists to Perform Downstream Pathway Enrichment Analysis and Target Prioritization

Published on: October 3, 2025

842
DNA-Tethered RNA Polymerase for Programmable In vitro Transcription and Molecular Computation
09:26

DNA-Tethered RNA Polymerase for Programmable In vitro Transcription and Molecular Computation

Published on: December 29, 2021

4.7K

Related Experiment Videos

Last Updated: Jan 2, 2026

Bidirectional Retroviral Integration Site PCR Methodology and Quantitative Data Analysis Workflow
12:53

Bidirectional Retroviral Integration Site PCR Methodology and Quantitative Data Analysis Workflow

Published on: June 14, 2017

11.1K
Using Human Differentially Expressed Gene Lists to Perform Downstream Pathway Enrichment Analysis and Target Prioritization
03:08

Using Human Differentially Expressed Gene Lists to Perform Downstream Pathway Enrichment Analysis and Target Prioritization

Published on: October 3, 2025

842
DNA-Tethered RNA Polymerase for Programmable In vitro Transcription and Molecular Computation
09:26

DNA-Tethered RNA Polymerase for Programmable In vitro Transcription and Molecular Computation

Published on: December 29, 2021

4.7K

Area of Science:

  • Genomics
  • Computational Biology
  • Bioinformatics

Background:

  • Neural networks achieve high performance in genome sequence prediction.
  • Interpreting learned features from these models, especially motifs, remains a significant challenge.

Purpose of the Study:

  • To explore and compare existing visualization techniques for inferring sequence information learned by recurrent neural networks (RNNs).
  • To assess the effectiveness of these techniques for splice junction identification in genomic sequences.

Main Methods:

  • Modulated existing visualization techniques for genome sequence input.
  • Applied perturbation-based and back-propagation-based methods to inspect genomic regions at nucleotide and span levels.
  • Utilized a recurrent neural network (RNN) for splice junction identification.

Main Results:

  • Visualization techniques successfully inferred both canonical and non-canonical splicing features from a single neural model.
  • Perturbation-based visualizations outperformed back-propagation-based ones for canonical splice motifs.
  • Back-propagation-based visualizations were superior for non-canonical splice motifs.
  • Comparable performance was observed for branchpoint detection between visualization methods.

Conclusions:

  • Visualization techniques are crucial for understanding neural network models in genomics.
  • The choice of visualization method impacts the accuracy of motif identification based on motif type (canonical vs. non-canonical).
  • The SpliceVisuL tool offers a practical solution for analyzing neural network-derived genomic sequence information.