关于 opentrons-integration
opentrons-integration 是用于为 Opentrons Flex 和 OT-2 实验室自动化机器人编写 Python 协议的技能。内容涵盖液体处理、实验室器材与模块控制、常见移液流程,以及执行前的协议模拟。
主要功能
- 01编写 Protocol API v2 协议
- 02配置移液器、器材和台面布局
- 03自动化液体处理操作
- 04控制温控、磁力及其他模块
- 05机器人执行前模拟协议
适用场景
- 01创建系列稀释协议
- 02搭建 PCR 工作流
- 03在微孔板间复制样本
- 04通过模拟测试移液协议
安装
复制下面的命令,在终端中运行。
$
npx skills add https:/ / github.com/ jackspace/ claudeskillz --skill opentrons-integration安装后在 Claude Code 中输入/opentrons-integration确认已加载。
附带文件SKILL.json
详细介绍
Opentrons 集成
概览
Opentrons 是一个基于 Python 的实验室自动化平台,适用于 Flex 和 OT-2 机器人。可编写用于液体处理的 Protocol API v2 协议、控制硬件模块(加热振荡器、热循环仪),以及管理实验室器皿,从而实现自动化移液工作流。
何时使用此技能
在以下情况下应使用此技能:
- 使用 Python 编写 Opentrons Protocol API v2 协议
- 在 Flex 或 OT-2 机器人上自动化液体处理工作流
- 控制硬件模块(温控、磁力、加热振荡器、热循环仪)
- 设置实验室器皿配置和工作台布局
- 实现复杂的移液操作(系列稀释、板复制、PCR 配置)
- 管理吸头使用并优化协议效率
- 使用多通道移液器进行 96 孔板操作
- 在机器人执行协议前进行模拟和测试
核心功能
1. 协议结构和元数据
每个 Opentrons 协议都遵循标准结构:
from opentrons import protocol_api
# Metadata
metadata = {
'protocolName': 'My Protocol',
'author': 'Name <[email protected]>',
'description': 'Protocol description',
'apiLevel': '2.19' # Use latest available API version
}
# Requirements (optional)
requirements = {
'robotType': 'Flex', # or 'OT-2'
'apiLevel': '2.19'
}
# Run function
def run(protocol: protocol_api.ProtocolContext):
# Protocol commands go here
pass
关键要素:
- 从
opentrons导入protocol_api - 定义包含 protocolName、author、description、apiLevel 的
metadata字典 - 可选的
requirements字典,用于指定机器人类型和 API 版本 - 实现
run()函数,将ProtocolContext作为参数接收 - 所有协议逻辑都放在
run()函数中
2. 加载硬件
加载仪器(移液器):
def run(protocol: protocol_api.ProtocolContext):
# Load pipette on specific mount
left_pipette = protocol.load_instrument(
'p1000_single_flex', # Instrument name
'left', # Mount: 'left' or 'right'
tip_racks=[tip_rack] # List of tip rack labware objects
)
常见的移液器名称:
- Flex:
p50_single_flex、p1000_single_flex、p50_multi_flex、p1000_multi_flex - OT-2:
p20_single_gen2、p300_single_gen2、p1000_single_gen2、p20_multi_gen2、p300_multi_gen2
加载实验室器皿:
# Load labware directly on deck
plate = protocol.load_labware(
'corning_96_wellplate_360ul_flat', # Labware API name
'D1', # Deck slot (Flex: A1-D3, OT-2: 1-11)
label='Sample Plate' # Optional display label
)
# Load tip rack
tip_rack = protocol.load_labware('opentrons_flex_96_tiprack_1000ul', 'C1')
# Load labware on adapter
adapter = protocol.load_adapter('opentrons_flex_96_tiprack_adapter', 'B1')
tips = adapter.load_labware('opentrons_flex_96_tiprack_200ul')
加载模块:
# Temperature module
temp_module = protocol.load_module('temperature module gen2', 'D3')
temp_plate = temp_module.load_labware('corning_96_wellplate_360ul_flat')
# Magnetic module
mag_module = protocol.load_module('magnetic module gen2', 'C2')
mag_plate = mag_module.load_labware('nest_96_wellplate_100ul_pcr_full_skirt')
# Heater-Shaker module
hs_module = protocol.load_module('heaterShakerModuleV1', 'D1')
hs_plate = hs_module.load_labware('corning_96_wellplate_360ul_flat')
# Thermocycler module (takes up specific slots automatically)
tc_module = protocol.load_module('thermocyclerModuleV2')
tc_plate = tc_module.load_labware('nest_96_wellplate_100ul_pcr_full_skirt')
3. 液体处理操作
基本操作:
# Pick up tip
pipette.pick_up_tip()
# Aspirate (draw liquid in)
pipette.aspirate(
volume=100, # Volume in µL
location=source['A1'] # Well or location object
)
# Dispense (expel liquid)
pipette.dispense(
volume=100,
location=dest['B1']
)
# Drop tip
pipette.drop_tip()
# Return tip to rack
pipette.return_tip()
复杂操作:
# Transfer (combines pick_up, aspirate, dispense, drop_tip)
pipette.transfer(
volume=100,
source=source_plate['A1'],
dest=dest_plate['B1'],
new_tip='always' # 'always', 'once', or 'never'
)
# Distribute (one source to multiple destinations)
pipette.distribute(
volume=50,
source=reservoir['A1'],
dest=[plate['A1'], plate['A2'], plate['A3']],
new_tip='once'
)
# Consolidate (multiple sources to one destination)
pipette.consolidate(
volume=50,
source=[plate['A1'], plate['A2'], plate['A3']],
dest=reservoir['A1'],
new_tip='once'
)
高级技巧:
# Mix (aspirate and dispense in same location)
pipette.mix(
repetitions=3,
volume=50,
location=plate['A1']
)
# Air gap (prevent dripping)
pipette.aspirate(100, source['A1'])
pipette.air_gap(20) # 20µL air gap
pipette.dispense(120, dest['A1'])
# Blow out (expel remaining liquid)
pipette.blow_out(location=dest['A1'].top())
# Touch tip (remove droplets on tip exterior)
pipette.touch_tip(location=plate['A1'])
流速控制:
# Set flow rates (µL/s)
pipette.flow_rate.aspirate = 150
pipette.flow_rate.dispense = 300
pipette.flow_rate.blow_out = 400
4. 访问孔和位置
孔访问方法:
# By name
well_a1 = plate['A1']
# By index
first_well = plate.wells()[0]
# All wells
all_wells = plate.wells() # Returns list
# By rows
rows = plate.rows() # Returns list of lists
row_a = plate.rows()[0] # All wells in row A
# By columns
columns = plate.columns() # Returns list of lists
column_1 = plate.columns()[0] # All wells in column 1
# Wells by name (dictionary)
wells_dict = plate.wells_by_name() # {'A1': Well, 'A2': Well, ...}
位置方法:
# Top of well (default: 1mm below top)
pipette.aspirate(100, well.top())
pipette.aspirate(100, well.top(z=5)) # 5mm above top
# Bottom of well (default: 1mm above bottom)
pipette.aspirate(100, well.bottom())
pipette.aspirate(100, well.bottom(z=2)) # 2mm above bottom
# Center of well
pipette.aspirate(100, well.center())
5. 硬件模块控制
温控模块:
# Set temperature
temp_module.set_temperature(celsius=4)
# Wait for temperature
temp_module.await_temperature(celsius=4)
# Deactivate
temp_module.deactivate()
# Check status
current_temp = temp_module.temperature # Current temperature
target_temp = temp_module.target # Target temperature
磁力模块:
# Engage (raise magnets)
mag_module.engage(height_from_base=10) # mm from labware base
# Disengage (lower magnets)
mag_module.disengage()
# Check status
is_engaged = mag_module.status # 'engaged' or 'disengaged'
加热振荡器模块:
# Set temperature
hs_module.set_target_temperature(celsius=37)
# Wait for temperature
hs_module.wait_for_temperature()
# Set shake speed
hs_module.set_and_wait_for_shake_speed(rpm=500)
# Close labware latch
hs_module.close_labware_latch()
# Open labware latch
hs_module.open_labware_latch()
# Deactivate heater
hs_module.deactivate_heater()
# Deactivate shaker
hs_module.deactivate_shaker()
热循环仪模块:
# Open lid
tc_module.open_lid()
# Close lid
tc_module.close_lid()
# Set lid temperature
tc_module.set_lid_temperature(celsius=105)
# Set block temperature
tc_module.set_block_temperature(
temperature=95,
hold_time_seconds=30,
hold_time_minutes=0.5,
block_max_volume=50 # µL per well
)
# Execute profile (PCR cycling)
profile = [
{'temperature': 95, 'hold_time_seconds': 30},
{'temperature': 57, 'hold_time_seconds': 30},
{'temperature': 72, 'hold_time_seconds': 60}
]
tc_module.execute_profile(
steps=profile,
repetitions=30,
block_max_volume=50
)
# Deactivate
tc_module.deactivate_lid()
tc_module.deactivate_block()
吸光度板读仪:
# Initialize and read
result = plate_reader.read(wavelengths=[450, 650])
# Access readings
absorbance_data = result # Dict with wavelength keys
6. 液体跟踪和标记
定义液体:
# Define liquid types
water = protocol.define_liquid(
name='Water',
description='Ultrapure water',
display_color='#0000FF' # Hex color code
)
sample = protocol.define_liquid(
name='Sample',
description='Cell lysate sample',
display_color='#FF0000'
)
将液体加载到孔中:
# Load liquid into specific wells
reservoir['A1'].load_liquid(liquid=water, volume=50000) # µL
plate['A1'].load_liquid(liquid=sample, volume=100)
# Mark wells as empty
plate['B1'].load_empty()
7. 协议控制和实用工具
执行控制:
# Pause protocol
protocol.pause(msg='Replace tip box and resume')
# Delay
protocol.delay(seconds=60)
protocol.delay(minutes=5)
# Comment (appears in logs)
protocol.comment('Starting serial dilution')
# Home robot
protocol.home()
条件逻辑:
# Check if simulating
if protocol.is_simulating():
protocol.comment('Running in simulation mode')
else:
protocol.comment('Running on actual robot')
导轨灯(仅限 Flex):
# Turn lights on
protocol.set_rail_lights(on=True)
# Turn lights off
protocol.set_rail_lights(on=False)
8. 多通道和 8 通道移液
使用多通道移液器时:
# Load 8-channel pipette
multi_pipette = protocol.load_instrument(
'p300_multi_gen2',
'left',
tip_racks=[tips]
)
# Access entire column with single well reference
multi_pipette.transfer(
volume=100,
source=source_plate['A1'], # Accesses entire column 1
dest=dest_plate['A1'] # Dispenses to entire column 1
)
# Use rows() for row-wise operations
for row in plate.rows():
multi_pipette.transfer(100, reservoir['A1'], row[0])
9. 常见协议模式
系列稀释:
def run(protocol: protocol_api.ProtocolContext):
# Load labware
tips = protocol.load_labware('opentrons_flex_96_tiprack_200ul', 'D1')
reservoir = protocol.load_labware('nest_12_reservoir_15ml', 'D2')
plate = protocol.load_labware('corning_96_wellplate_360ul_flat', 'D3')
# Load pipette
p300 = protocol.load_instrument('p300_single_flex', 'left', tip_racks=[tips])
# Add diluent to all wells except first
p300.transfer(100, reservoir['A1'], plate.rows()[0][1:])
# Serial dilution across row
p300.transfer(
100,
plate.rows()[0][:11], # Source: wells 0-10
plate.rows()[0][1:], # Dest: wells 1-11
mix_after=(3, 50), # Mix 3x with 50µL after dispense
new_tip='always'
)
板复制:
def run(protocol: protocol_api.ProtocolContext):
# Load labware
tips = protocol.load_labware('opentrons_flex_96_tiprack_1000ul', 'C1')
source = protocol.load_labware('corning_96_wellplate_360ul_flat', 'D1')
dest = protocol.load_labware('corning_96_wellplate_360ul_flat', 'D2')
# Load pipette
p1000 = protocol.load_instrument('p1000_single_flex', 'left', tip_racks=[tips])
# Transfer from all wells in source to dest
p1000.transfer(
100,
source.wells(),
dest.wells(),
new_tip='always'
)
PCR 配置:
def run(protocol: protocol_api.ProtocolContext):
# Load thermocycler
tc_mod = protocol.load_module('thermocyclerModuleV2')
tc_plate = tc_mod.load_labware('nest_96_wellplate_100ul_pcr_full_skirt')
# Load tips and reagents
tips = protocol.load_labware('opentrons_flex_96_tiprack_200ul', 'C1')
reagents = protocol.load_labware('opentrons_24_tuberack_nest_1.5ml_snapcap', 'D1')
# Load pipette
p300 = protocol.load_instrument('p300_single_flex', 'left', tip_racks=[tips])
# Open thermocycler lid
tc_mod.open_lid()
# Distribute master mix
p300.distribute(
20,
reagents['A1'],
tc_plate.wells(),
new_tip='once'
)
# Add samples (example for first 8 wells)
for i, well in enumerate(tc_plate.wells()[:8]):
p300.transfer(5, reagents.wells()[i+1], well, new_tip='always')
# Run PCR
tc_mod.close_lid()
tc_mod.set_lid_temperature(105)
# PCR profile
tc_mod.set_block_temperature(95, hold_time_seconds=180)
profile = [
{'temperature': 95, 'hold_time_seconds': 15},
{'temperature': 60, 'hold_time_seconds': 30},
{'temperature': 72, 'hold_time_seconds': 30}
]
tc_mod.execute_profile(steps=profile, repetitions=35, block_max_volume=25)
tc_mod.set_block_temperature(72, hold_time_minutes=5)
tc_mod.set_block_temperature(4)
tc_mod.deactivate_lid()
tc_mod.open_lid()
最佳实践
- 始终指定 API 级别:在元数据中使用最新的稳定 API 版本
- 使用有意义的标签:为实验室器皿添加标签,便于在日志中识别
- 检查吸头是否充足:确保吸头数量足以完成协议
- 添加注释:使用
protocol.comment()进行调试和记录日志 - 先进行模拟:在机器人上运行前,始终先模拟测试协议
- 妥善处理错误:在需要时添加暂停,以便手动干预
- 考虑时间因素:协议需要孵育时使用延迟
- 跟踪液体:使用液体跟踪来更好地验证实验设置
- 优化吸头使用:适当时使用
new_tip='once'以节省吸头 - 控制流速:针对黏稠或易挥发液体调整流速
故障排除
常见问题:
- 吸头用尽:确认吸头盒容量符合协议要求
- 实验室器皿碰撞:检查工作台布局,确认不存在空间冲突
- 体积错误:确保液体体积不超过孔或移液器的容量
- 模块无响应:确认模块连接正常且固件已更新
- 体积不准确:校准移液器并检查是否有气泡
- 协议在模拟中失败:检查 API 版本兼容性和实验室器皿定义
资源
有关 API 的详细文档,请参阅此技能目录中的 references/api_reference.md。
有关协议模板示例,请参阅 scripts/ 目录。
评论
支持 Markdown