Use Targets to compare candidate targets for a disease and plan research into gaps in their evidence.
Create a prioritization
- Open Targets and select New prioritization.
- Enter the Disease and any Target symbols you want to include.
- Open Optional details to add an Ontology ID, imported scorecard or own-data filename. Upload any named files in the project's Data tab after creating it.
- Optional details also contains Cohort cap, which limits the number of targets, and GTEx tissue.
- Select Create project.
- Review the new project, upload any named files, and start research.
Creating the project does not start a run. Until results are available, Targets shows Not built yet, with a link to the project. Search the prioritizations list by name or disease to find existing work.
If the project is created but its inputs cannot be saved, use Retry in the dialog. Open project lets you inspect the project already created.
Read the ranking
Open a prioritization from Targets. The header shows the snapshot date; open Provenance and version for the artifact version, timestamp and source metadata.
- Search by target symbol, synonym or full name. Select a column heading to sort.
- Open Filters to narrow the list; selected filters stay visible and can be cleared individually.
- Select a target to inspect its score, coverage, direction, rank sensitivity and evidence in the adjacent reader. On smaller screens, use Back to ranking to return to the list.
- Open Evidence matrix for every channel, expandable measures and their sort controls. A blank value means the target was not assessed for that measure.
Search, filters and selection survive a trip to a full profile, Compare or Proposals within the prioritization. Returning brings the relevant target back into view. List and matrix each retain their own sort so switching views does not silently sort by a hidden evidence column.
Coverage helps you judge how much evidence supports a score. Rank range shows how the ranking changes when the channel weights vary. A narrow range indicates a more stable ranking under those changes; it does not establish that a target will succeed.
Leverage identifies a missing or weak evidence channel that could change a target's rank. Use it to decide which gaps to investigate.
Change the weights
Select Weights. Switch channels or individual measures on or off, then adjust their weights. The ranking updates using the existing evidence.
Weights are saved in your browser for that prioritization version. If browser storage is unavailable, the page tells you that the changes will not persist. Use Reset to return to the default weights.
The Provenance filter changes which evidence columns are visible. It does not change the scoring weights.
Compare and export
Open Compare to compare targets across the available measures.
In the list, Export ranking downloads the visible targets in their current order, including names, rank-range and leverage fields. In Evidence matrix, Export matrix also includes the displayed evidence channels and expanded measures. Missing values remain blank.
Read a dossier
Select a target, then Full profile, to inspect its dossier. Research proposals, Evidence and Linked projects near the top jump to those sections. Review source dates, missing evidence and linked project results before choosing follow-up work.
- Find papers opens an Atlas search for the target and disease.
- Discuss opens the prioritization project with the dossier as chat context.
If a refresh fails, the page retains the last loaded result and shows a warning. Use Retry before treating it as current.
Act on a proposal
Open Proposals to review suggested work on evidence gaps. Search by target,
question, project or disease name, or gap. Combine terms such as LRRK2 genetics,
then use the status filters and cost or potential-rank-change sort. Clear search
removes the query without resetting the selected status.
- Create project opens a review before anything is saved. Edit the project name and mission; open Project details to review its context, goals and opening chat message. The fixed scientific brief travels with those details, including the question, method, data, expected outcome and source. Confirm to create the linked project, or cancel without creating it. The plan and scientific brief are saved for the research run; the opening message appears unsent in its research chat. Start research separately from the project.
- Add to project saves the complete proposal as a brief in an existing project and queues it for Marvin to review. Choose a project of the proposal's type, or a full research project. The existing mission and goals stay intact. The confirmation names the queued proposal and offers Open project; queueing does not start research or mean Marvin has adopted the work.
- Dismiss removes a proposal from the open work list. Reopen makes it available again.
Linked to project records that handoff; it does not mean research is running. Open Research details for the proposed approach, data, expected outcome and goals. Potential rank change compares the target's rank if the gap's evidence matched the cohort median versus its maximum under your weights; it does not predict the study result.
Changing a proposal requires write access and an inactive prioritization run. Adding to an existing project also requires that project's run to be inactive.
If a handoff is unfinished, use its Retry or Open project action. Recovery remembers the accepted study and destination across refreshes in the same browser tab, unless the page reports that browser storage is unavailable. Retry completes the remaining steps; an unconfirmed queue request keeps its original identity so already-accepted work is not duplicated. If the proposal has changed or the destination is unavailable, the page explains why it cannot finish. Unfinished handoffs remain accessible in Proposals even if a later run removes the original proposal. Forgetting a handoff removes only this browser reminder; it does not delete a project or cancel queued work.
The project target link records its original prioritization. A proposal brief also carries its own source link, so adding another study for an already-declared target preserves both origins. Finishing a follow-up project does not automatically update the source ranking.