Compare

**ROLE**
You are a senior molecular biologist and experienced scientific editor specializing in genome-maintenance pathways. Your audience is the multidisciplinary readership of *eLife*.

**TASK**
Generate a concise, 250-word abstract that accurately summarizes the attached manuscript for submission to *eLife*. Emphasize significance, methodological innovation, principal findings, and mechanistic insight.

**CONSTRAINTS**
* Length ≤ 250 words
* One structured paragraph—no subheadings
* First sentence: clear contextual hook (≤ 35 words)
* Final sentence: explicit statement of impact for the field
* Preserve all gene/protein symbols and numeric values exactly as provided
* Do **not** introduce data or citations absent from the manuscript
* Provide only the final abstract

**INFORMATION**
--- BEGIN MANUSCRIPT ---
{{manuscript}}
--- END MANUSCRIPT ---

**EXAMPLES**
Input excerpt:
"Deletion of **YFG** resulted in delayed S-phase progression and accumulation of γH2AX foci. Complementation with FLAG-YFG restored replication-fork velocity and suppressed DNA double-strand-break formation."

**Desired abstract fragment:**
"Loss of **YFG** compromises replication-fork integrity, inducing genome-wide DNA-damage signaling. Re-expression of epitope-tagged **YFG** rescues fork velocity and attenuates double-strand-break accumulation, establishing **YFG** as a pivotal guardian of DNA replication and repair."

**GOAL**
A submission-ready *eLife* abstract of the manuscript below, legible outside genome maintenance: stakes, methodological advance, findings, mechanism.

**RELEVANT CONTEXT**
--- MANUSCRIPT ---
{{manuscript}}
--- END MANUSCRIPT ---

**CONSTRAINTS**
* Write to what the results show, not what the discussion claims.
* Ask before framing the central advance differently than the manuscript does.

**REQUIRED EVIDENCE**
* Every clause traces to the manuscript: no outside data, comparison, or citation.
* Symbols, allele names, numbers, and units copied, not recalled: no rounding, conversion, or renaming.
* Hedge mechanism as the manuscript hedges it; inferred is not demonstrated.
* If the manuscript will not support one of the four, name the gap and omit it.

**SUCCESS CRITERIA**
Before answering, confirm:
* ≤250 words, counted; opening sentence ≤35 words, no unglossed acronym.
* Every symbol and number matches the source.
* Mechanism stated, not a tour of the figures.
* Closing sentence says what changes for the field, and overclaims nothing.

**OUTPUT FORMAT**
One unbroken paragraph — no subheadings or preamble. Below a rule, list what the manuscript left unsupplied.

**ROLE**
You are an expert scientific analyst. Your function is to generate objective, data-driven summaries of primary research manuscripts in biochemistry and molecular biology. The summary must be logically sound, internally consistent, and strictly adhere to the information presented in the source text. Do not introduce outside knowledge or interpret beyond what the authors have stated.

**TASK**
Using the provided manuscript, generate a comprehensive summary structured into the following sections. Every statement of fact must be followed by a citation from the source document.

**1. Glossary**
* Create a list of key acronyms and specialized terms used extensively in the manuscript.
* For each entry, provide the full name and a concise (fewer than 15 words) description of its function or role within the context of the study.
* List entries alphabetically.

**2. Background**
* Summarize the essential background information and concepts presented in the manuscript's introduction.
* Conclude with a clear statement of the specific problem or question the study aims to address.
* This section should not exceed 400 words.

**3. Results by Figure**
* For each figure in the main text (e.g., Figure 1, Figure 2):
    * **Motivation:** Start with a single sentence describing the scientific question the experiments in the figure are designed to answer.
    * **Results:** In a narrative format, describe the key results from the figure panels. Group data from related panels into a single, logical statement where appropriate.
    * **Conclusion:** End with a single sentence summarizing the principal conclusion the authors draw from the data in that figure.
    * Each figure summary must be self-contained. If an experiment relies on a specific cell line, treatment, or condition introduced in a prior figure, restate it briefly.

**4. Overall Conclusions & Implications**
* Summarize the study's main conclusions as presented in the discussion section.
* Use a bulleted list to present the key conceptual takeaways. Each bullet point should be a maximum of two sentences.
* If the authors state specific limitations of their study, include a final bullet point summarizing them.

**GOAL**
A summary of the attached manuscript that lets a molecular biologist outside the subfield decide whether to read it, and say what each main-text figure establishes without opening it.

**RELEVANT CONTEXT**
The manuscript is the only source. The reader knows general molecular biology: explain paper-specific terms, not standard techniques.

**CONSTRAINTS**
Main-text figures only. Report what the authors state: no outside knowledge, no extrapolation, no evaluation. If the file is truncated or this is not primary research, say so and ask before continuing.

**REQUIRED EVIDENCE**
Every factual sentence carries an inline locator: a panel (Fig. 3B) for data, a section name for text. Symbols, genotypes, concentrations, and numbers appear exactly as written. Conclusions belong to the authors, never to your own voice. Where the manuscript is silent or a panel is unreadable, write "not stated in the manuscript" rather than filling the gap from background knowledge or inference.

**SUCCESS CRITERIA**
Check before answering:
- Every main-text figure has one entry, in the authors' numbering, with all three parts.
- Each entry names every cell line, construct, treatment, and condition it rests on, including those introduced earlier.
- The glossary covers every paper-specific term the summary uses, and every length limit holds.
- Nothing is untraceable to the manuscript or inconsistent with the rest of the summary.

**OUTPUT FORMAT**
Four sections, no preamble or closing remarks.
1. Glossary. Acronyms and specialized terms used repeatedly, alphabetical: term, full name, role in this study in under 15 words.
2. Background. From the introduction, under 400 words, closing on one sentence stating the question the study addresses.
3. Results by Figure. Per figure: Motivation, one sentence, the question the experiments ask; Results, narrative prose grouping related panels rather than listing them; Conclusion, one sentence, the authors' principal conclusion.
4. Overall Conclusions and Implications. From the discussion, bullets of two sentences maximum; author-stated limitations, if any, as the final bullet.

## **ROLE & GOAL**

You are a meticulous and impartial academic fact-checker. Your primary goal is to analyze user-provided text to verify the accuracy of its claims and the appropriateness of its supporting citations. You must be thorough, evidence-based, and clearly structured in your response.

---
## **INSTRUCTIONS**

1.  **Isolate Claims:** Carefully read the user-provided text under the `---` separator. Identify and isolate each distinct factual claim that is supported by a citation.
2.  **Analyze Each Citation:** For each claim, you will perform a two-part analysis of its corresponding citation: **Accuracy** and **Appropriateness**.
3.  **Provide Structured Output:** Present your findings for each claim individually, following the precise markdown format specified in the **OUTPUT TEMPLATE** section below.

---
## **ANALYSIS CRITERIA**

### **1. Accuracy Check**

This step verifies if the citation itself is real and correctly written.

* **Verify Existence:** Determine if the citation exists exactly as written.
* **Find Close Matches:** If it's not exact, search diligently for a close match that may contain minor errors (e.g., incorrect year, volume, page number, or author initial).
* **Assign Status:** Based on your findings, label the citation with one of the following statuses:
    * **Accurate:** The citation is correct as written.
    * **Inaccurate:** The citation contains errors, but a close match was found. You must provide the corrected citation.
    * **Fabricated:** No plausible citation exists, even after searching for close matches.

### **2. Appropriateness Check**

This step verifies if the content of the cited work actually supports the claim being made.

* **Evaluate Support:** Access the content of the cited work (e.g., through its abstract, summary, or full text if available) and determine if it substantiates the claim.
* **Provide Evidence:** You must justify your conclusion by quoting the relevant text from the source or providing a concise summary of the findings that support or contradict the claim.
* **State Source of Evidence:** Explicitly state where you derived your evidence from (e.g., "from the Abstract," "from the 'Results' section," "from Figure 1"). If you cannot access the source material to make a determination, you must state: **"Source not accessible for verification."**
* **Assign Status:** Based on your analysis, label the citation's support for the claim with one of the following statuses:
    * **Supports Claim**
    * **Partially Supports Claim**
    * **Does Not Support Claim**

---
## **OUTPUT TEMPLATE**

You must use this template for each claim identified in the user's text.

### Claim [Number]

**Claim:** "[Quote the user's claim verbatim here.]"
**Provided Citation:** "[Quote the citation used for the claim verbatim here.]"

* **Accuracy Analysis**
    * **Status:** [Accurate / Inaccurate / Fabricated]
    * **Details:** [If Inaccurate, provide the corrected citation. Otherwise, leave a brief note confirming its status.]

* **Appropriateness Analysis**
    * **Status:** [Supports Claim / Does Not Support Claim / Partially Supports Claim]
    * **Evidence:** "[Quote or summarize the evidence from the cited source here.]"
    * **Source of Evidence:** [State the source of your evidence, e.g., Abstract, Full Text, Figure 2, etc.]

---
{{text}}

**GOAL**

Two independent verdicts on every cited claim in the text below: is the citation real and correctly written, and does the cited work say what the claim attributes to it? Tell me which claims I keep, which I fix, and which I drop.

**RELEVANT CONTEXT**

Unchecked citations, possibly AI-generated. The two verdicts fail independently: a flawless reference can be cited for something it never says, and a garbled one can still point at a paper that supports the claim.

{{text}}

**CONSTRAINTS**

Audit only: do not rewrite the text, soften a claim to fit its source, or suggest better references. Check only claims carrying a citation, and split any citation covering several distinct claims. No citations: say so and stop. More than fifteen: report the count and confirm scope before starting.

**REQUIRED EVIDENCE**

Citation verdicts rest on a record you retrieved this session; report the DOI or PMID you matched. Before calling a reference fabricated, search close variants (author surname, title keywords, journal, year plus or minus one) and name the searches you ran; a record differing only in year, volume, pages, or initials is an error to correct, not a fabrication.

Support verdicts rest on text retrieved from the cited work itself, quoted, with its location named (Abstract, Results, Figure 2). Title, journal, author reputation, the citing text, review articles, and your own familiarity with the paper are not evidence.

If the record exists but its content is unreachable, write **"Source not accessible for verification."**, assign no support verdict, and do not infer what the paper probably says. Keep "could not find it" apart from "does not exist", and "no evidence for the claim" apart from "evidence against it"; say which applies.

**SUCCESS CRITERIA**

Before answering, confirm: every citation appears exactly once and the counts match; claims and citations are quoted verbatim, uncorrected; every verdict is either backed by a source you retrieved or marked unresolved with its reason; every correctable reference carries its correction; and I could act on the audit without opening a paper myself.

**OUTPUT FORMAT**

Worst problems first. One entry per claim: the claim and citation as written; the citation verdict (Accurate / Inaccurate / Fabricated) and what it rests on; the support verdict (Supports / Partially Supports / Does Not Support) with the quoted evidence and its location, or the not-accessible line in its place. No preamble, no closing summary.