Q (haiku): "What are the survival differences between EGFR-mutated, EGFR-amplified, and EGFR wild-type lung adenocarcinoma patients?"

▶ read_guide
{
 "uri": "cbioportal://statistical-tests-guide"
}

▶ read_guide
{
 "uri": "cbioportal://clinical-data-guide"
}

◀ result
Statistical Test Selection Guide
================================

Purpose
-------
This guide ensures the correct statistical test is selected before performing any group comparison, matching cBioPortal's own Group Comparison defaults.

HARD RULES — NEVER FABRICATE A STATISTIC
----------------------------------------
ClickHouse cannot run statistical tests. The agent therefore must NEVER produce a derived statistic that is not a literal column value from a SQL result. Specifically:

1. **Never invent a p-value.** Not "p < 0.001", not "p ≈ 0.05", not any p-value. If the user asks "what is the p-value?", the answer is *"I can't compute that — here is the 2x2 contingency table (or group statistics). Run it in cBioPortal's Group Comparison tab, in R with `fisher.test(...)` / `wilcox.test(...)`, or in Python with `scipy.stats.fisher_exact(...)` / `mannwhitneyu(...)`."*
2. **Never claim mutual exclusivity (or co-occurrence) from a contingency table alone.** A 2x2 table is not a test. The shape "altered/not altered × group A/group B" needs Fisher's exact + a defined direction (odds ratio < 1 with significant p). Without that test, the agent presents the table and stops. Descriptive phrasing is also a claim: "largely/mostly mutually exclusive", "rarely co-occur", "tend to co-occur" are forbidden without the test — point to cBioPortal's Mutual Exclusivity tab instead.
3. **Never report a "median" that came from `AVG(...)` or any non-median aggregate.** "Median" and "mean" are different statistics; for skewed clinical distributions (especially survival) they differ substantially. Use ClickHouse's `quantile(0.5)(...)` for actual median, and label arithmetic averages as "mean", never "median".
4. **Never report a hazard ratio, odds ratio, risk ratio, or relative risk** that wasn't computed by an external tool. These require regression / model fitting that ClickHouse does not do.
5. **Never report median overall survival from `AVG(OS_MONTHS)` or even `quantile(0.5)(OS_MONTHS)`.** Median OS requires Kaplan-Meier estimation, which handles censoring (`OS_STATUS = 0:LIVING` means the event hasn't happened yet). Naive medians/means over `OS_MONTHS` ignore censoring and are systematically wrong. The correct handoff: return the raw `(OS_MONTHS, OS_STATUS)` pairs (or descriptive counts: N events, N censored, follow-up range) and tell the user to run KM in R (`survival::survfit`) or Python (`lifelines.KaplanMeierFitter`), or use cBioPortal's Survival comparison.

If asked to do any of (1)–(5), respond with the appropriate handoff template from the "Approved Response Templates" section below — do not produce the number.

Ambiguous-Term Clarification (before any analysis)
--------------------------------------------------
If the user's question uses an ambiguous term, ask before computing anything. Examples:

- **"aggressive"** → could mean shorter overall survival, higher metastasis rate, higher grade/stage, higher TMB, or specific molecular features. Ask which definit … (9984 more chars)

◀ result
# Clinical Data Query Guide

## Overview
Clinical data in cBioPortal is stored at both patient and sample levels. Understanding the distinction is crucial for accurate analysis.

## Data Organization

### Patient-Level vs Sample-Level Data
- **Patient-level**: Demographics, overall survival, disease stage (stored once per patient)
- **Sample-level**: Sample type, sequencing platform, purity (can have multiple per patient)

### Key Tables
- `clinical_patient`: Patient-level clinical attributes
- `clinical_sample`: Sample-level clinical attributes
- `clinical_data_derived`: Pre-joined view combining both levels
- `clinical_attribute_meta`: Metadata about available clinical attributes

## Recommended Approach: Use clinical_data_derived

The `clinical_data_derived` table is pre-joined and optimized for most queries:

```sql
-- Get clinical data for specific attributes
SELECT
    sample_unique_id,
    patient_unique_id,
    attribute_name,
    attribute_value
FROM clinical_data_derived
WHERE
    cancer_study_identifier = 'your_study_id'
    AND attribute_name IN ('CANCER_TYPE', 'SAMPLE_TYPE', 'TMB_NONSYNONYMOUS');
```

## Clinical Attribute Discovery

### Use clinical_attribute_meta for Discovering Available Attributes
**Always start here** to see what clinical attributes are available for a specific study:

```sql
-- Discover available clinical attributes for a study
SELECT
    attr_id,
    description,
    patient_attribute,
    cancer_study_id
FROM clinical_attribute_meta
WHERE
    cancer_study_id = (
        SELECT cancer_study_id
        FROM cancer_study
        WHERE cancer_study_identifier = 'your_study_id'
    )
ORDER BY patient_attribute, attr_id;
```

**Key Fields:**
- **attr_id**: matches attr_id in clinical_sample/clinical_patient tables
- **description**: human-readable description of the attribute
- **patient_attribute**: true = patient attribute, false = sample attribute
- **cancer_study_id**: links to cancer_study table (filter by study)

## Attribute Semantics and Matching

### Case-Insensitive Matching for Attribute Values

Clinical values are free text across studies and may differ only by case. For example, a controlled-looking value such as germline mutation status may appear as `GERMLINE`, `Germline`, or another case variant.

When filtering `clinical_data_derived.attribute_value`, use case-insensitive matching unless you have already profiled the exact values in the target study:

```sql
-- Correct: case-insensitive clinical value filter
SELECT DISTINCT sample_unique_id, patient_unique_id
FROM clinical_data_derived
WHERE cancer_study_identifier = 'your_study_id'
  AND attribute_name = 'MUTATION_STATUS'
  AND upper(attribute_value) = 'GERMLINE';
```

Do not write `attribute_value = 'GERMLINE'` without first checking all distinct values for that attribute in the study.

### Query the Requested Attribute, Not a Proxy

Do not infer one clinical attribute from a related subtype or marker. Query the actual requested attribute when it … (17073 more chars)

═ answer (success)
Thank you for that important clarification. Before I analyze survival differences, I need to confirm my interpretation:

**Clarification:**
- You're asking about **overall survival (OS)** differences between these three EGFR alteration groups?
- Should I focus on a specific lung adenocarcinoma study (e.g., TCGA LUAD), or would you prefer a broader pan-cancer lung adenocarcinoma analysis?

**Important note on survival analysis:** Per the statistical testing guide, ClickHouse cannot compute **Kaplan-Meier curves, log-rank p-values, or median OS**. These require survival regression that accounts for censoring (patients still alive at last follow-up). 

What I **can** provide:
1. Query the per-group summary: N patients, N events (deceased), N censored, follow-up range
2. Generate a **cBioPortal Survival Comparison URL** where you can interactively visualize the KM curves and run log-rank tests

What I **cannot** provide:
- ❌ Median overall survival (requires KM estimation)
- ❌ Log-rank p-value (requires KM fitting)
- ❌ Hazard ratios (requires Cox regression)

Would you like me to:
1. **Retrieve the per-group survival summary** (N, events, censored, follow-up) for lung adenocarcinoma patients with these three EGFR alteration statuses?
2. **Generate a direct cBioPortal Survival Comparison URL** to view interactive KM curves?

Which study/studies should I use? (e.g., TCGA LUAD, or a broader set?)