grants

---
name: grants
description: Use when the user asks about NIH research funding — finding grants, positioning a research idea, or matching it to institutes/mechanisms (e.g. 'grants for X', 'find NIH funding').
category: Research
version: 1.0.0
tools: []
---

# Grants — NIH Funding Intelligence

Produce a strategic NIH funding overview for a clinical researcher's research idea, delivered as a structured chat report covering positioning, institute mapping, targeted grant discovery, and strategic recommendations the researcher can copy into a proposal or share with a mentor.

**Scope: NIH only.** Non-NIH funders (PCORI, DOD CDMRP, VA, foundations) are out of scope and should be flagged at intake.

## Integrity Rules

- Search sequentially — don't fire overlapping queries.
- Count only what searches actually returned this session; never supplement with unlabeled background knowledge. Label anything not sourced this session `[Not from search — reference information]` and exclude it from citation counts.
- Track three separate numbers: queries sent / results shown / results cited. Every cited paper or project needs a retrievable URL.
- On failure, retry once with a reworded query; after repeated failures, stop and tell the researcher what's missing rather than guessing.
- Keep an audit-log section in the final report with the same standards as the body.

## Phase 1: Intake (6 questions, one at a time)

1. **Research idea** in 2-3 sentences — the question, what's new, the clinical relevance. Reject vague answers ("AI for healthcare," "biomarkers for disease X") and push for specificity; a vague idea produces vague positioning language and useless gap analysis.
2. **Career stage** — pre-doctoral (PhD/T32), postdoctoral (F32/K99), early career (K-award/first R01), independent investigator (multiple R01s), or senior PI (R35/P-series/U01). This filters which mechanisms are even sane to recommend (F-series for trainees, K-series for early career, R-series for independent).
3. **Preliminary data status** — none (de novo), pilot (early single-site), strong preliminary (multi-experiment, R01-scale ready), or validated (multi-site, publication-ready). Drives mechanism scope: no data points to R03/R21 pilot scope, strong prelim supports R01/U01 multi-site scale.
4. **Research environment** — R01-eligible institution, mid-tier academic medical center, resource-constrained institution, or industry-collaborative. Affects scope realism (multi-site U01 needs an R01-eligible base) and which mechanisms are competitive (R15 specifically targets resource-constrained institutions).
5. **Submission posture** — new application, resubmission (A1, needs reviewer-response guidance), or still exploring.
6. **Known institute targets** — specific NIH institutes already in mind (NCI, NHLBI, NIMH, NINDS, NIDDK, etc.), or "no preference — find the right ones." If they have a hypothesis, validate it against the funding data; if not, surface the top three institutes funding adjacent work.

Stop after question 6 and commit — never reopen intake mid-run.

## Phase 2A: Research Positioning (5 searches)

Run five searches, one per positioning facet, and extract 2-3 quotable findings from each for the report's gap-quote section:

1. **Established** — what's already known about the topic.
2. **Stakes** — mortality, burden, cost, or prevalence data showing why it matters.
3. **Current approaches** — standard of care / current treatment landscape.
4. **Adjacent methods** — related techniques applied to this topic.
5. **Gaps** — limitations, unanswered questions, named future directions.

Draft Significance/Innovation language using the pattern: "the field has established X (refs), but Y remains unanswered (refs)."

## Phase 2B: Institute Mapping + Grant Discovery

Search NIH RePORTER (reporter.nih.gov) for funded projects matching the research idea, using both a narrow AND-style query (direct-overlap terms) and a broader OR-style query with synonyms (adjacent work). Use a rolling window of the current fiscal year plus the three prior (federal FY starts October 1).

From the results:
- **Tally funding institutes** (agency/IC code — NCI, NHLBI, NIMH, etc.) to rank the top 3.
- **Tally study sections** to rank the top 2 — these are where the application will actually be reviewed.
- **Look for NOSI opportunity numbers** (format `NOT-<INSTITUTE>-<YEAR>-<NUMBER>`, published at grants.nih.gov/grants/guide/notice-files/) among the funded-project results; fetch each one found. If a fetch fails, note it and continue rather than fabricating detail.

## Mechanism Matching

Mechanism recommendation is never career-stage alone — it's **career stage × project scope × preliminary-data status**. A trainee with strong multi-site preliminary data and an R01-eligible environment shortlists differently than a trainee with none. Use the intake answers (questions 2, 3, 4, plus intended scope) together to shortlist mechanisms and give the rationale for each.

## Report Structure

1. **Executive Summary** — title, career stage, environment, 3-4 key findings.
2. **Research Positioning** — 3-5 gap quotes (cited) plus a 2-3 paragraph positioning narrative.
3. **Target Institutes** — ranking table (institute, project count in window, rough % match) plus interpretation.
4. **Grant Opportunities** — any NOSI called out prominently; top-3 grants with links plus a scope/budget-fit paragraph per grant.
5. **Funded Overlap** — top-5 comparable funded projects (PI, project number, institute, year, link) plus a differentiation paragraph explaining how this idea is distinct.
6. **Study Sections** — ranking plus interpretation of the best match.
7. **Strategic Recommendations** — 3-4 numbered recommendations, the mandatory program-officer recommendation (below), a submission-timeline note, reviewer-response guidance if this is a resubmission, and a closing paragraph.
8. **References** — numbered, hyperlinked bibliography.
9. **Audit Log** — searches run, counts, any failed steps.

### Mandatory Program Officer Recommendation

Always include in the recommendations section: **contact the program officer at the top-ranked institute.** Point the researcher to the NIH institute staff directory (nih.gov/institutes-nih/list-nih-institutes-centers-offices → institute → Program Officers) and tell them to prepare a one-page specific-aims summary, their CV, and three specific fit questions before reaching out. This is the single most valuable piece of advice for any applicant — never skip it.

### Submission Timeline

| Mechanism | Standard receipt dates |
|---|---|
| R01, R21, R03 | Feb 5, Jun 5, Oct 5 |
| K awards (K01, K08, K23, K99) | Feb 12, Jun 12, Oct 12 |
| R34, R61/R33 | Feb 16, Jun 16, Oct 16 |
| F31, F32 | Apr 8, Aug 8, Dec 8 |

## Anti-Patterns

- Recommending a mechanism from career stage alone, ignoring scope and preliminary data.
- Filling thin search results with unlabeled background knowledge instead of surfacing the gap explicitly.
- Skipping the audit log or the program-officer recommendation.
- Conflating "results shown" with "results cited."
- Fabricating NOSI detail when a fetch fails — log it and move on.

## Error Handling

| Failure | Response |
|---|---|
| A search returns nothing for a facet | State that explicitly; don't backfill with training knowledge. |
| RePORTER search returns very few results on the narrow pass | Note it; the broad pass should compensate. |
| NOSI fetch fails | Log `[NOSI — fetch failed]` and continue. |
| Repeated failures across searches | Stop, tell the researcher what's missing, share what was found. |