Add research gap replication planner

research-gap-replication-planner/README.md

@@ -0,0 +1,70 @@
+# Research Gap Replication Planner
+
+Self-contained contribution for `SCIBASE-AI/SCIBASE.AI#16`, focused on the boundary between the AI Research Assistant Suite's peer-review, reproducibility, and research-gap capabilities.
+
+This slice is intentionally not another broad assistant demo. It builds a deterministic planner that turns corpus limitations, prior reproducibility attempts, manuscript evidence coverage, and lab capabilities into reviewer-ready replication priorities.
+
+## What It Does
+
+- Generates auto peer-review flags for claim/evidence/citation alignment, statistical risk, and domain-specific omissions.
+- Scores project reproducibility readiness from raw data, clean pipelines, environment locks, tests, reported outputs, and previous run attempts.
+- Finds under-replicated research intersections with unresolved findings, negative-result signals, and lab-fit evidence.
+- Produces a prioritized replication plan with required artifacts, first actions, and a stable audit digest.
+- Uses only local synthetic data and Node.js standard library APIs.
+
+## Requirement Map
+
+| Issue #16 requirement | Implementation evidence |
+| --- | --- |
+| Auto peer review reports | `reviewManuscript` emits structured review findings with severity, category, claim id, and domain-specific template checks. |
+| Claims vs. evidence alignment | Claims without evidence, citations, or registered support are flagged before release. |
+| Statistical or methodological red flags | Small-sample p-value edge cases, missing controls, and missing domain evidence are detected. |
+| Reproducibility checker | `scoreReproducibility` checks required artifacts, prior attempts, environment integrity, and output consistency. |
+| Reproducibility confidence score | Each project receives a numeric confidence score plus pass/hold status and remediation actions. |
+| Links to previous reproducibility attempts | Prior attempt ids and outcomes are included in each report and used in scoring. |
+| Research gap finder | `rankResearchOpportunities` ranks under-replicated topic intersections using unresolved findings, negative signals, replication counts, and corpus activity. |
+| User opportunity feed | Lab capabilities and researcher interests shape the replication-priority output. |
+
+See `docs/requirement-map.md` for the detailed checklist.
+
+## Demo
+
+```bash
+cd research-gap-replication-planner
+npm run demo
+```
+
+Expected summary:
+
+```text
+Planner status: hold
+Top opportunity: alzheimer + crispr + single-cell
+Top replication decision: prioritize
+Top peer-review flag: Claim has no supporting evidence artifact.
+Weak reproducibility project: circadian-metabolomics-2026
+Audit hash: de1b1fa44644dc24...
+```
+
+Visual demo assets:
+
+- `docs/demo.svg`
+- `docs/demo.mp4`
+
+## Verification
+
+```bash
+cd research-gap-replication-planner
+npm run check
+```
+
+Local verification covers:
+
+- opportunity ranking
+- peer-review flag generation
+- reproducibility scoring
+- deterministic audit hash generation
+- CLI demo output
+
+## AI-Assisted Disclosure
+
+This contribution was produced with AI assistance and reviewed before submission. It uses no credentials, no live user data, and no external services.

research-gap-replication-planner/docs/requirement-map.md

@@ -0,0 +1,13 @@
+# Requirement Map
+
+| Requirement | Evidence in this module | Verification |
+| --- | --- | --- |
+| Natural-language review suggestions | `src/planner.js` creates reviewer action text for unsupported claims, missing citations, small samples, and domain-specific omissions. | `test/planner.test.js` checks the highest severity review finding. |
+| Clarity and coherence checks | Review packets flag overbroad claims that lack an evidence artifact or citation trail. | `npm test` validates unsupported-claim detection. |
+| Statistical or methodological red flags | Small sample sizes near significance thresholds and missing domain controls are flagged. | `npm test` checks statistical risk categories. |
+| Claims vs. evidence alignment | Every claim is checked for evidence ids, citation ids, protocol linkage, and method support. | `npm test` checks claim alignment output. |
+| Reproducibility confidence | Projects are scored from raw data, clean pipeline, environment lock, tests, reported output, and prior attempts. | `npm test` verifies low-confidence and high-confidence cases. |
+| Previous reproducibility attempts | Attempt records are attached to project reports and affect the score. | `npm test` verifies failed attempts reduce confidence. |
+| Research gap finder | Corpus findings are grouped by topic intersection and scored for unresolved status, replication count, corpus activity, lab fit, and feasibility. | `npm test` verifies the top ranked topic intersection. |
+| Personalized opportunity feed | Lab capabilities and researcher interests are used to shape priority and first actions. | `npm run demo` prints the ranked plan and audit hash. |
+| Reviewer-ready evidence | Output includes required artifacts, first actions, gap rationale, and audit digest. | `npm run check` runs tests plus demo. |

research-gap-replication-planner/package.json

@@ -0,0 +1,12 @@
+{
+  "name": "research-gap-replication-planner",
+  "version": "0.1.0",
+  "description": "Deterministic research gap and replication-priority planner for SCIBASE issue #16.",
+  "type": "module",
+  "scripts": {
+    "demo": "node scripts/demo.js",
+    "test": "node test/planner.test.js",
+    "check": "npm test && npm run demo"
+  },
+  "license": "MIT"
+}

research-gap-replication-planner/sample/corpus.json

@@ -0,0 +1,109 @@
+{
+  "lab": {
+    "name": "Neurogenomics Replication Lab",
+    "interests": ["Alzheimer", "CRISPR", "single-cell", "metabolomics"],
+    "capabilities": ["single-cell", "RNA-seq", "CRISPR perturbation", "metabolomics"],
+    "methods": ["notebook rerun", "containerized workflow"]
+  },
+  "manuscripts": [
+    {
+      "id": "ms-neuro-042",
+      "title": "Perturbation response in single-cell Alzheimer models",
+      "domain": "molecular-biology",
+      "claims": [
+        {
+          "id": "c1",
+          "text": "CRISPR perturbation improves cell viability across all donor lines.",
+          "sampleSize": 18,
+          "pValue": 0.051,
+          "evidenceIds": [],
+          "citationIds": ["doi:10.1000/example-a"]
+        },
+        {
+          "id": "c2",
+          "text": "A subtype-specific response appears in microglia-like cells.",
+          "sampleSize": 64,
+          "pValue": 0.02,
+          "evidenceIds": ["notebook:microglia-response"],
+          "citationIds": []
+        }
+      ]
+    }
+  ],
+  "projects": [
+    {
+      "id": "circadian-metabolomics-2026",
+      "title": "Circadian metabolomics benchmark",
+      "dependencyIntegrity": "stale",
+      "outputConsistency": "drifted",
+      "artifacts": {
+        "rawData": true,
+        "cleanPipeline": true,
+        "environmentLock": false,
+        "testSet": false,
+        "reportedResults": true
+      }
+    },
+    {
+      "id": "microglia-rnaseq-2026",
+      "title": "Microglia RNA-seq replication packet",
+      "dependencyIntegrity": "locked",
+      "outputConsistency": "consistent",
+      "artifacts": {
+        "rawData": true,
+        "cleanPipeline": true,
+        "environmentLock": true,
+        "testSet": true,
+        "reportedResults": true
+      }
+    }
+  ],
+  "reproducibilityAttempts": [
+    {
+      "id": "run-031",
+      "projectId": "circadian-metabolomics-2026",
+      "topicId": "finding-crispr-ad",
+      "outcome": "failed",
+      "note": "Pipeline required an unavailable R package version."
+    },
+    {
+      "id": "run-044",
+      "projectId": "microglia-rnaseq-2026",
+      "topicId": "finding-crispr-ad",
+      "outcome": "passed",
+      "note": "Notebook reran with matching summary metrics."
+    }
+  ],
+  "corpusFindings": [
+    {
+      "id": "finding-crispr-ad",
+      "title": "CRISPR perturbation in Alzheimer's single-cell models",
+      "topics": ["CRISPR", "Alzheimer", "single-cell"],
+      "methods": ["RNA-seq", "CRISPR perturbation"],
+      "citations": 126,
+      "activeProjects": 4,
+      "replicationCount": 1,
+      "limitation": "Unresolved donor variability and low replication across single-cell CRISPR Alzheimer studies."
+    },
+    {
+      "id": "finding-sleep-metabolomics",
+      "title": "Sleep disruption metabolomics in aging cohorts",
+      "topics": ["sleep", "metabolomics", "aging"],
+      "methods": ["metabolomics"],
+      "citations": 38,
+      "activeProjects": 2,
+      "replicationCount": 3,
+      "limitation": "Negative results were rarely reported and methods varied across cohorts."
+    },
+    {
+      "id": "finding-protein-folding",
+      "title": "Protein folding benchmark coverage",
+      "topics": ["protein folding", "benchmark"],
+      "methods": ["simulation"],
+      "citations": 14,
+      "activeProjects": 1,
+      "replicationCount": 5,
+      "limitation": "Benchmark appears stable with several successful replications."
+    }
+  ]
+}

research-gap-replication-planner/scripts/demo.js

@@ -0,0 +1,21 @@
+import fs from "node:fs"
+import path from "node:path"
+import { fileURLToPath } from "node:url"
+import { createResearchAssistantPlan } from "../src/planner.js"
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url))
+const samplePath = path.join(__dirname, "..", "sample", "corpus.json")
+const sample = JSON.parse(fs.readFileSync(samplePath, "utf8"))
+const plan = createResearchAssistantPlan(sample)
+
+const topOpportunity = plan.opportunityFeed[0]
+const topPlan = plan.replicationPlans[0]
+const topFinding = plan.peerReviewReports[0].findings[0]
+const weakProject = plan.reproducibilityReports.find((report) => report.status !== "ready")
+
+console.log(`Planner status: ${plan.status}`)
+console.log(`Top opportunity: ${topOpportunity.topic}`)
+console.log(`Top replication decision: ${topPlan.decision}`)
+console.log(`Top peer-review flag: ${topFinding.message}`)
+console.log(`Weak reproducibility project: ${weakProject.projectId}`)
+console.log(`Audit hash: ${plan.auditHash.slice(0, 16)}...`)