Review Fusobacterium protocol counted as original target and conflicting outcome codes

The 14 September 2026 explorer snapshot links replication `10.7554/elife.25801` to both:

- `10.1101/gr.126516.111`, *Fusobacterium nucleatum infection is prevalent in human colorectal carcinoma*: **mixed**.
- `10.7554/elife.10012`, *Registered report: Fusobacterium nucleatum infection is prevalent in human colorectal carcinoma*: **failed**.

The [Registered Report PDF](https://cdn.elifesciences.org/elife-articles/10012/pdf/elife-10012.pdf) describes the plan to replicate Castellarin et al. The second relation appears to connect protocol and results, rather than a second original empirical finding. It risks inflating target/outcome counts and misrepresenting the protocol as a failed original.

Please verify relation types and the intended coding unit against the [replication report](https://doi.org/10.7554/elife.25801), reconcile mixed/failed labels, preserve protocol provenance separately where appropriate, and refresh downstream summaries. Do not deduplicate indiscriminately by replication DOI: legitimate reports can test multiple originals.

Start by comparing the replication report at 10.7554/elife.25801 with the Registered Report PDF for 10.7554/elife.10012 and the linked original at 10.1101/gr.126516.111. Trace how relation types and outcome codes feed downstream summaries; done means protocol provenance is distinct, mixed/failed labels are reconciled, and summaries are refreshed without collapsing legitimate multi-target reports.

Written by the indexing model from the issue text.