Don't have an account?
Enter the username or email associated with your account and we'll send you a link to reset your password.
Database of Ebola Virus Integrated Screens
DEVISe
Database of Ebola Virus Integrated Screens
Search Ebola virus–host protein interactions from integrated functional genomic screens, explore protein complexes, and generate interaction networks.
Published data is available without login. Access to unpublished data is restricted to members of NIH P01 AI120943. Contact devise@bu.edu with questions.
Database of Ebola Virus Integrated Screens
Curated protein complexes containing your searched genes, from Complex Portal, opens in a new tab (EMBL-EBI).
The network is an interactive diagram explored with a mouse or touch: drag to pan, scroll to zoom, and select a node or edge for details. It is not keyboard navigable. The same interactions, scores and study references are available as sortable, screen-reader accessible text in the Gene Lookup and Protein Complexes tabs, and can be downloaded with Download results or Cytoscape file.
The DEVISe scoring algorithm combines evidence strength and diversity to assess the evidence that a protein is a functional Ebola virus interactor. The formula integrates multiple data types with empirically determined weights.
Survival scores indicating a gene's essentiality (from Wang et al.) are included in database to inform users about the feasibility of conducting gene knockout experiments and are not considered when calculating DEVISe score.
In words: the score equals alpha times one minus the exponential of the negative of the sum over data type t, manuscript m and data instance i of the experiment-type weight k for t, m and i multiplied by the ECDF s raised to the power gamma; plus beta times the number of data types with at least one piece of evidence, divided by D-max. Each parameter is defined in the two tables that follow.
| Fixed Hyperparameters | Value | Description |
|---|---|---|
| α | 0.9152563 | Weight of evidence strength parameter |
| β | 0.0847437 | Weight of evidence diversity parameter |
| ktranscriptomic | 1.85015 | Effective weight of transcriptomic data class, scaled with other k values to map evidence strength from 0–1 |
| kgenetic | 1.972419 | Weight of genetic data class, scaled with other k values to map evidence strength from 0–1 |
| kproximity | 0.02612437 | Weight of proximity data class, scaled with other k values to map evidence strength from 0–1 |
| kphysical | 1.373895 | Weight of physical data class, scaled with other k values to map evidence strength from 0–1 |
| γ | 0.001 | Scaling variable for importance of s relative to k |
| Dmax | 4 | Number of independent data classes in DEVISe database which can contribute to DEVISe score |
| Variable parameters | Description |
|---|---|
| Score | DEVISe evidence score that a protein is a functional Ebola virus interactor |
| t | Index of data type: proximity, protein interaction, genetic, transcriptomic |
| m | Index of manuscript containing evidence for a gene |
| i | Index of specific data instance for a particular protein |
| s | Empiric cumulative density function (ECDF), indicating how highly a piece of evidence ranks within the experiment in which it was found |
How DEVISe scores are spread across every gene in the database. Genes are split into four equal-sized confidence levels by score quartile, and each level is shaded in the curve. Search for genes in the left-hand column to mark where they fall, or show the known positives, genes with established roles in Ebola virus infection, to see where they score.
The plot is a density curve of DEVISe scores, lowest on the left and highest on the right, shaded by confidence level, with a band under it marking where each level starts and ends. Each marked gene is a button: tab to it to hear its score, confidence level and rank, and press Enter to open it in Gene Lookup, or, where several genes share one spot, to list them all. Every value in the plot is also in the table view below it.
How DEVISe evidence flows from experimental class, to publication, to gene. Search for genes in the left-hand column, hover any box or ribbon to trace its connections, select a class or publication to focus on it, and select a gene to see its supporting studies. Widths are to scale: every box is as tall, and every ribbon as wide, as the number of gene–study links it carries, so a study that contributed many genes is visibly heavier than one that contributed a few.
Each box in the diagram is a button. Tab to a class or publication box and press Enter to focus the diagram on it; press Enter on a gene box to list its supporting studies. Every box is as tall, and every ribbon as wide, as the number of gene–study links it carries, so the three columns are drawn to the same scale.
The Database of Ebola Virus Integrated Screens (DEVISe) is a comprehensive resource combining data from multiple functional genomics screens to identify host factors involved in Ebola virus infection.
This database integrates diverse experimental evidence including proximity-dependent biotinylation, protein–protein interaction studies, genetic screens, and transcriptomic data to compute an evidence score for each protein's role in the viral lifecycle.
Funding
This work is supported by the National Institutes of Health under grant P01 AI120943.
Site creator: Jillian Grassia, PhD candidate, Davey Lab | LinkedIn
Technical contributor: Ben Slutzky, NEIDL IT Operations Leader | LinkedIn
Advisor: Robert Davey, Professor | LinkedIn
DEVISe's application architecture and full-stack development were assisted by Claude (Anthropic).
For questions, feedback, or data inquiries, contact the DEVISe team at devise@bu.edu.