OmevraOmevra · Tools
Tools
Every usable tool in one directory — what each one does, and where it runs.
Variant Evidence Agent — tools
These run in the live application, over the scored universe and current public records. Open the application and choose the tab named here.
Variant report
One variant in — a cited evidence summary with its calibrated probability out, or an explicit decline where the evidence is not there. The Report tab of the application; the full system is described on the Variant Evidence Agent page.
Batch evidence pack
Many variants at once: paste a list (or a single gene name to expand it), or upload a variant-list file, and receive one evidence pack — a verdict table, the per-variant reports, and both downloadable packs. The Batch tab of the application.
Variant normaliser
Paste a variant in any spelling — the full transcript name, the colon form, gene and variant, variant alone, or an allele id — and see the cleaned form the agent uses, its status in the scored universe, and its calibrated probability when it has one. One parser serves this tool and the Report tab, so the two can never disagree. The Tools tab of the application.
Gene explorer
Open one gene: how many of its variants sit in the scored universe, how many the agent summarises and how many it declines, its published per-gene calibration where one exists, and the variants themselves — select any of them to open its report. The Tools tab of the application.
Literature panel (PubMed)
Part of live-evidence mode: alongside the current ClinVar and gnomAD records, the agent retrieves the variant's PubMed literature — title, journal, year and a linked PMID — as retrieved citations. Retrieval only: no score is attached, and nothing is claimed about what a paper concludes; when the lookup cannot be completed, the panel says so.
Recent batches
Below the Batch tab, completed batches are listed with their counts and one-click re-download of their packs. The list is saved on your own device, in your browser — it is not an account and nothing is kept on a server.
web-agent — tools
Live run
Pick a task, press Run, and watch the agent work in a real browser — observe, ground, act, verify, and recover when the page fights back — with the results panel and a step-by-step trace. Run the same task with recovery off to see the measured difference. The full results are on the web-agent page.
Public-page mode
Point the agent at a public web page with one of two fixed intents — find and click the element with a label, or type into the field with a label and submit — inside hard safety rails: public addresses only, no password or payment fields, no leaving the entered domain, a small step budget. The rails are described on the architecture page.
On this site
Ask Omevra
The assistant at the bottom-left of every page on this site. It answers questions about the products, the methods and the results from this site's own pages only, names the page each answer comes from, and declines plainly when a question is outside what the site documents.
Downloads — the raw result files
The web-agent result set (Evaluation v1.0) ships with this site: the recovery-on and recovery-off run files and the generated results document, byte-identical to the lab's own copies, linked from the web-agent page.
Methods library
How the results were produced, in the projects' own terms: evaluation, calibration, and temporal evaluation.
Variant Evidence Agent — API
The live application is also a public JSON API: variant lookup, search, gene counts, single reports and batch evidence packs, with enforced bounds. Endpoints, limits and copyable examples with real responses: API documentation.