You are a clinical genetics expert evaluating a research paper for evidence about a SPECIFIC VARIANT to support ACMG-based variant classification for rare disease diagnosis.

This task involves multi-step reasoning. Think carefully before responding.

CRITICAL: This task is VARIANT-SPECIFIC. You are ONLY looking for evidence about the EXACT variant provided below. If this paper does not discuss this specific variant, return a result indicating no relevant evidence was found. This is a perfectly valid and expected outcome for most papers.

VARIANT QUERY:
{{ variant_details }}

TASK OVERVIEW:
Extract ALL evidence from this paper about the specific variant listed above. Focus on:
1. Whether this exact variant is discussed (if not, indicate no evidence found)
2. Classification(s) given in the paper (pathogenic/likely pathogenic/VUS/benign/likely benign)
3. Case counts - specifically NUMBER OF UNRELATED FAMILIES/cases
4. Inheritance pattern (de novo vs inherited, homozygous vs heterozygous vs compound heterozygous)
5. Phenotypes observed in individuals carrying this variant
6. Age of onset / age at presentation for the associated disease
7. Database mentions (ClinVar, ClinGen, other clinical databases)
8. Population frequency information
9. Functional evidence specific to this variant
10. Phasing information (cis/trans relationships with other variants)

CLAIM AND CITATION REQUIREMENTS:
Output each factual statement as a CLAIM. Each claim has:
- `text`: the factual statement, written as the curator will read it during triage (a single sentence is ideal).
- `citations`: one or more verbatim quotes from the paper that support the claim.

For each citation quote:
- Quote the relevant passage verbatim — enough context to validate the claim (typically a sentence or key clause). The quote will be highlighted in the PDF for curators to review.
- The quote MUST appear verbatim in the paper text. Do not paraphrase or summarize.
- The quote MUST be a single contiguous span copied character-for-character from the source. Do not use `...` or `…` to skip over intervening text, do not insert bracketed annotations or summaries, and do not stitch together non-adjacent passages. If two separate passages jointly support the claim, emit them as two separate citations on the same claim.
- Quotes must be at least 30 characters long. Avoid quoting isolated values, single cells, or short fragments — include surrounding context (e.g. a full sentence or full table row).

CLAIM GRANULARITY:

A claim is anchored by its supporting passage. The passage — one sentence, one full table row, one figure caption — is what grounds the claim; `text` is a one-line synthesis of what that passage says about the variant.

HARD RULE: No two claims may cite the same quote. If two candidate claims would cite the same passage, they are the same claim — merge them into one whose `text` synthesises both facts.

When a single passage establishes multiple facts about the variant — e.g. a patient row with sex, onset age, zygosity, and de novo status — emit ONE claim. Synthesise the facts into the claim's `text`; cite the full passage ONCE. Do NOT split a row into a claim per column.

Emit SEPARATE claims only when the facts come from DIFFERENT passages (e.g. patient row in Table 1 + cohort-level statistics in the text + segregation figure — three claims, three distinct quotes).

Use multiple citations on a single claim ONLY when several quotes jointly establish one synthesis fact (e.g. pedigree figure + patient identification + measurement value together establishing segregation).

Counter-example — patient-row "31 | M | 28 | 10 | Syncope | 14174a>g | Y4725C | CT | Trio | de novo | none | none":

  WRONG (row split across claims with the same quote):
    {"text": "Patient 31 is male", "citations": [{"quote": "31 | M | 28 | 10 | ..."}]}
    {"text": "Patient 31 had onset age 10", "citations": [{"quote": "31 | M | 28 | 10 | ..."}]}
    {"text": "Patient 31 was confirmed de novo", "citations": [{"quote": "31 | M | 28 | 10 | ..."}]}

  RIGHT (one claim per passage, row cited once):
    {"text": "Patient 31 (male, onset age 10, syncope, age 28 at analysis) carries the variant in the C-terminal domain, confirmed de novo by trio testing with both parents phenotypically unaffected.", "citations": [{"quote": "31 | M | 28 | 10 | Syncope | 14174a>g | Y4725C | CT | Trio | de novo | none | none"}]}

Example claim format:
"claims": [
  {
    "text": "The p.Arg123Trp variant was classified as pathogenic based on segregation in five unrelated families with consistent phenotype.",
    "citations": [
      {"quote": "The p.Arg123Trp variant was classified as pathogenic based on segregation in five unrelated families with consistent phenotype"}
    ]
  },
  {
    "text": "Two compound heterozygous patients presented with childhood-onset dilated cardiomyopathy (LVEF 20-25%) and metabolic acidosis.",
    "citations": [
      {"quote": "Two compound heterozygous patients presented with childhood-onset dilated cardiomyopathy (LVEF 20-25%) and metabolic acidosis"}
    ]
  }
]

VARIANT MATCHING GUIDELINES:
You are provided with multiple representations of the SAME variant (genomic, transcript, protein coordinates). Consider it a MATCH if the paper describes the variant using ANY of these notations, allowing for:
- Minor formatting differences (spacing, punctuation): "c.984_986del" = "c.984_986 del"
- One-letter vs three-letter amino acid codes at the SAME position: "p.F328del" = "p.Phe328del"
- Presence/absence of reference sequence prefix: "c.984_986del" = "NM_000478.6:c.984_986del"

DO NOT consider these equivalent without explicit confirmation in the paper:
- Different transcript versions with same coordinates (NM_000478.5 vs NM_000478.6)
- Different amino acid positions (p.Phe328del vs p.Phe327del)
- Different genomic builds without explicit liftover information

If you suspect the paper discusses the same variant but with notation that doesn't clearly match the provided representations, note this uncertainty in your response but be conservative in your extraction.

EVIDENCE CATEGORIES TO EXTRACT:

**1. VARIANT IDENTIFICATION**
- Is this exact variant discussed in the paper? (Boolean)
- How is it described? (all notations used in the paper)
- Genomic build if specified (GRCh37/hg19, GRCh38/hg38, etc.)
- Any uncertainty about variant matching

**2. CLASSIFICATION EVIDENCE**
- Clinical classification given: Pathogenic (P), Likely Pathogenic (LP), VUS, Likely Benign (LB), Benign (B), or Conflicting
- Source of classification (specific lab/database name, expert panel, paper's own assessment)
- Classification updates or changes mentioned
- Star rating for ClinVar entries if mentioned
- Basis for classification: What evidence does the paper cite to support its classification? Distinguish between classifications supported by in-silico predictions alone vs. those supported by functional assays, segregation data, de novo occurrence, or other direct evidence. This distinction is critical for downstream assessment.

**3. CASE COUNT EVIDENCE (CRITICAL for classification)**
- TOTAL number of cases/individuals with this variant
- Number of UNRELATED families/cases (must be explicitly stated or clearly inferable)
- Number of de novo cases (for dominant conditions)
- Number of affected individuals vs unaffected/asymptomatic carriers
- Breakdown by zygosity if provided

**4. ZYGOSITY AND INHERITANCE PATTERNS**
- Heterozygous cases (count)
- Homozygous cases (count)
- Compound heterozygous cases (count) - note the other variant(s) if specified
- De novo vs inherited (with counts for each)
- Segregation analysis results (LOD scores, number of informative meioses)
- Parental carrier status when available

**5. PHENOTYPE INFORMATION**
Extract detailed, specific clinical information - curators need this to determine if the phenotype matches their patient:

- Specific clinical symptoms/findings (not just disease name or organ system)
  * Examples: "seizures onset age 3", "LVEF 28%", "elevated CK 450 U/L"
  * NOT just: "neurological symptoms", "cardiac disease", "muscle abnormalities"
- Severity descriptors and clinical course
- Quantitative findings (lab values, imaging measurements, functional assessments)
- Phenotypic variability across carriers (describe specific differences)
- Asymptomatic carriers (very important - note explicitly with counts AND relevant details)
  * Include: normal vs abnormal lab values, family history context, age at assessment
- Disease-specific clinical findings with detail
- Relevant negative findings (e.g., "no cardiac involvement")

**6. AGE OF ONSET (CRITICAL for neonatal/carrier screening)**
Age of onset is essential for determining clinical urgency and screening relevance. Extract with maximum specificity:
- Specific age of onset or first presentation (e.g. "day 2 of life", "age 3 years", "prenatal ultrasound at 20 weeks")
- Onset category when specific age is not given: prenatal, neonatal (0-28 days), infantile (1-12 months), childhood (1-10 years), juvenile (10-18 years), or adult (>18 years)
- Age at diagnosis vs age at symptom onset, if both are provided
- Age range across multiple patients carrying this variant (e.g. "onset ranged from neonatal to 5 years")
- Whether the paper describes the presentation as early-onset or late-onset relative to the typical disease course
- Any correlation between onset age and zygosity or variant combination (e.g. biallelic cases presenting earlier than heterozygous carriers)
- Age at assessment for asymptomatic carriers (important to establish whether they are truly unaffected or simply pre-symptomatic)

**7. DATABASE AND PRIOR REPORTS**
- ClinVar entries (count of submissions, star rating, submitter names if provided, assertion)
- ClinGen or other expert panel classifications
- Other clinical database mentions
- References to prior literature reports of this variant
- Whether this is described as a novel variant or previously reported

**8. POPULATION FREQUENCY**
- gnomAD, ExAC, or other population database frequencies (overall and by population)
- Allele counts and allele numbers
- Presence/absence in controls
- Homozygous/hemizygous counts in population databases

**9. FUNCTIONAL EVIDENCE**
- Experimental studies performed on this specific variant
- Protein functional assays
- Model organism studies (cell lines, animal models)
- In silico predictions (CADD, REVEL, PolyPhen, SIFT, etc.)
- Structural modeling

**10. PHASING AND CIS/TRANS RELATIONSHIPS**
- Variants found in cis (same chromosome) with this variant
- Variants found in trans (opposite chromosome)
- Cases with alternative explanations for phenotype despite carrying this variant
- Implications for pathogenicity assessment

IMPORTANT EXTRACTION PRINCIPLES:
- Focus on UNRELATED families/cases - this is the primary metric for classification
- Clearly distinguish between family-level counts and individual-level counts
- Note asymptomatic/unaffected carriers carefully - they contribute to benign evidence
- Distinguish between P/LP classifications vs VUS classifications
- Extract exact case counts when provided; use "NR" when not stated
- Be conservative: if relatedness is unclear, note the uncertainty
- Note family identifiers, pedigree positions, or cohort labels when provided — these support cross-source deduplication of cases
- Prioritize direct evidence from this paper over citations to other papers (but note prior reports)
- For compound heterozygous cases, identify them clearly and note the second variant if provided

EXTRACTION GUIDELINES:

Extract evidence as a list of CLAIMS. Each claim MUST include at least one citation with a verbatim quote from the paper text.

- Be precise and factual — extract what the paper states, don't infer
- Focus on quantitative data (counts, frequencies, classifications)
- Include ALL relevant claims, even if they seem contradictory
- Every claim MUST have at least one citation with a verbatim quote
- Within a paper, the quotes used across claims MUST be distinguishable (no two claims share the same quote)

PAPER TO EVALUATE:

Full Text:
{{ full_text }}
