Biotek Gen5

Offline mode works immediately with built-in demo datasets. API and GUI modes require Windows with Gen5 software.

Note: API and GUI modes are defined but not yet implemented. connect() returns False for both.

Quick Start

# Note: Directory is 'biotek-gen5' on disk but imports use underscores
from device_skills.schema import ControlMode
from devices.biotek_gen5.adapter import Gen5Adapter

reader = Gen5Adapter(mode=ControlMode.OFFLINE)
reader.connect()

# List available datasets
datasets = reader.list_datasets()
# Returns: [{"name": "ELISA_IL6_plate1", "mode": "absorbance", ...}, ...]

# Process a dataset
reading = reader.process("ELISA_IL6_plate1")
print(f"Mode: {reading.mode.value}, Wavelength: {reading.wavelength_nm}nm")
print(f"Well A1: OD={reading.plate.get_well('A1').value}")

reader.disconnect()

Key Workflows

1. Run Assay and Read Plate

In OFFLINE mode, acquire() is not available (no physical reader). Use process() to load data from demo datasets or exported files.

reader = Gen5Adapter(mode=ControlMode.OFFLINE)
reader.connect()

# ELISA absorbance data
elisa = reader.process("ELISA_IL6_plate1")

# Cell viability fluorescence data (name matching: "viability" or "cell")
viability = reader.process("CellViability_DrugScreen")

Dataset name matching: names containing "viability" or "cell" return fluorescence data; all others return absorbance data.

2. Process Exported Data

WARNING: In OFFLINE mode, adapter.process() returns demo data for testing, not real file contents. Use Gen5Processor directly for real CSV files.

# For real exported CSV files, use the processor directly:
from devices.biotek_gen5.processor import Gen5Processor
processor = Gen5Processor()
raw = processor.load("/path/to/export.csv")
corrected = processor.transform(raw)
reading = processor.extract(corrected)

3. Process CSV Exports (Detailed Pipeline)

Use Gen5Processor for the three-step pipeline on Gen5 CSV files:

from devices.biotek_gen5.processor import Gen5Processor

processor = Gen5Processor()

# Step 1: Parse CSV (metadata header rows + 8x12 grid)
raw = processor.load("data/elisa_plate1.csv")
# Returns: {"wells": {"A1": 0.123, ...}, "metadata": {...}, "file": "..."}

# Step 2: Blank correction (cols 11-12 as blanks)
corrected = processor.transform(raw)
# Adds: "blank_avg" and "blank_corrected" dict

# Step 3: Extract structured PlateReading
result = processor.extract(corrected)
reading = result["reading"]   # PlateReading object
stats = result["stats"]       # {"well_count", "min", "max", "mean", "blank_avg"}

CSV format expected: Gen5 export with metadata rows at top, then rows A-H with 12 numeric columns. Non-row lines are parsed as key-value metadata.

4. Analyze Results with AI

from devices.biotek_gen5.brain import Gen5Brain

brain = Gen5Brain()  # Uses ANTHROPIC_API_KEY; falls back to cached responses

# From adapter output
reading = reader.process("ELISA_IL6_plate1")
analysis = brain.analyze({"reading": reading})

# Streaming analysis
for chunk in brain.interpret_reading(reading, stream=True):
    print(chunk, end="")

# With experimental context
analysis = brain.interpret_reading(
    reading,
    context="Testing IL-6 response to LPS stimulation in THP-1 cells",
)

# Summarize multiple findings
summary = brain.summarize([analysis_1, analysis_2])

Without ANTHROPIC_API_KEY, the brain serves cached responses for ELISA and cell viability protocols.

5. Visualize Plate Data

from devices.biotek_gen5.visualizer import plot_plate_heatmap

# Generate heatmap PNG
png_bytes = plot_plate_heatmap(reading, output_path="output/plate.png")

# Bytes only (no file save)
png_bytes = plot_plate_heatmap(reading, output_path=None, title="My ELISA")

Colormap: YlOrRd for absorbance, YlGn for fluorescence. Wells are annotated with values.

6. Export to CSV

csv_string = Gen5Adapter.reading_to_csv(reading)
# Gen5-style CSV: header row with protocol/mode/wavelength, then 8x12 grid

7. File-Based Driver (for labclaw integration)

from pathlib import Path
from devices.biotek_gen5.driver import Gen5Driver

driver = Gen5Driver(device_id="gen5-001", watch_path=Path("/data/gen5/"))
await driver.connect()

# Reads most recently modified CSV in watch_path
data = await driver.read()
# Returns: {"wells": {"A1": 0.123, ...}, "metadata": {...}, "file": "...", "device_id": "gen5-001"}

# Driver is read-only; write() always returns False
await driver.disconnect()

Data Models

Demo Datasets

ELISA demo layout: Standards in cols 1-2 (serial dilution A-H), samples in cols 3-10, blanks in cols 11-12.

Cell viability demo layout: Positive controls (healthy cells) in cols 1-2, drug treatment in cols 3-4 (dose-dependent by row), test compounds in cols 5-10, negative controls (dead cells) in cols 11-12.

Plate Layout Conventions

• Columns 1-2: Standards or positive controls
• Columns 3-10: Samples or test compounds
• Columns 11-12: Blanks or negative controls
• Edge wells (row A/H, col 1/12): Prone to evaporation artifacts in long incubations (>2h at 37C)

Important Notes

• CSV parsing: Gen5Processor expects the specific Gen5 export format, not generic CSVs
• Blank correction: Automatic in both adapter (demo data) and processor pipeline, using cols 11-12
• Quality thresholds: CV% >15% = warning, >25% = problematic; Z-factor >0.5 = good screening assay

Reference Documentation