### 4-1. Quick Start

The previously used `google-generativeai` library has been deprecated. Therefore, you should install and start using the latest Google standard, the unified `google-genai` SDK.

```bash
# Install the latest unified Google Gen AI Python SDK
pip install google-genai
```

After installation, obtain an API key and set it as an environment variable.
```bash
export GEMINI_API_KEY="your_api_key"
```

This is an example Python script that performs a basic text query.
```python
from google import genai

# When creating a client, it automatically loads the GEMINI_API_KEY environment variable.
client = genai.Client()

response = client.models.generate_content(
    model='gemini-2.5-flash',
    contents='Explain the key differences between the CRISPR-Cas9 system and the CRISPR-Cas12a system.'
)

print(response.text)
```

### 4-2. Detailed Installation

If you need to systematically manage credits and pipelines in an enterprise environment or within the Google Cloud Platform (Vertex AI), it is recommended to configure and run it by linking the `gcloud` CLI as shown below.

```bash
# Log in to the gcloud CLI and generate an application authentication token
gcloud auth application-default login
```

This is a detailed code example for parsing a large number of documents for research purposes and accurately returning them in a structured data format (JSON).

```python
from google import genai
from google.genai import types

client = genai.Client()

prompt = """
Read the following PubMed abstract and return the target patient group, analyzed biomarkers,
and significantly changed values in a JSON format.
"Abstract: In this cohort of 120 patients with non-small cell lung cancer, we analyzed the expression of EGFR and KRAS.
We observed a 2.5-fold increase in EGFR mRNA levels in responders compared to non-responders..."
"""

# Specify the JSON Schema to enforce a structured response format
response = client.models.generate_content(
    model='gemini-2.5-pro',
    contents=prompt,
    config=types.GenerateContentConfig(
        response_mime_type="application/json",
        response_schema={
            "type": "OBJECT",
            "properties": {
                "patient_group": {"type": "STRING"},
                "biomarkers": {
                    "type": "ARRAY",
                    "items": {"type": "STRING"}
                },
                "expression_change": {"type": "STRING"}
            },
            "required": ["patient_group", "biomarkers", "expression_change"]
        }
    )
)

print(response.text)
```