retrobiosciences/opentrons-10x

OT-2 10x scRNA seq library prep (with multiplexing!)

---

• This protocol is difficult to get working properly
• The OT-2 p20 pipette tips are not always straight, making magnet separation especially difficult with stock p20 tips
• Please contact ben at ben@retro.bio before attempting this for yourself

---

Manual single cell library preps are tedious and error-prone. Automated solutions cost >$260k. This repo contains all code & hardware required to automate 10x Genomics library preps on Opentrons for $26k.

Table of Contents

• Protocol
• Results
• Required modifications
• Modules
• Labware
• Pipettes
• Reagents
• Optional
• Deck Setup
• Economic Efficiency
• Library prep quality

Protocol

Prepping 8 libraries involves ~2 hours of human labor¹.

Step 1: GEM Generation & Barcoding

• Perform manually using chromium controller
• Prepare reagents, plates, and labware in Opentrons

Step 2: (1:30)

• Pipette GEMs into PCR plate, add pink separation agent, wait 2 minutes, remove separation agent
• Begin automated portion
• Return within 24 hours to begin Step 3

Step 3: (4:30-5:00)

• Quantify DNA ²
  • update thermo-cycles-3.5.json with appropriate PCR cycle count
• If multiplexing:
  • reload pipette tips
• Begin automated portion
• Return within 24 hours and retrieve prepared sample (held at 4C indefinitely)
• Submit sample for sequencing!

Results

To showcase the performance of the automated protocol we generated multiple QC plots.

Number of reads

Automated	Manual

Quality scores

Automated	Manual

scVI integration

scVI is the de-facto standard method for correcting technical batch effects in single-cell data. Here, we applied it using manual/automated as our batch variables and plotted the resulting integration. If nothing but technical noise distinguishes our two samples, we expect to see perfect overlap.

Required modifications

• Opentrons Superior Magnet Module

Modules

• Thermocycler Module
• Magnetic Module
• Temperature Module

Labware

• 96-well aluminum block
• Bio-Rad Hard-Shell 96-Well PCR Plate, high profile, semi skirted #HSS9601 (2x)
• NEST 0.1ul PCR plate full-skirt
• Opentrons 300ul Tips (4-6x)
• Opentrons 20ul Tips (2-3x)
• 12-well reagent trough

Pipettes

• P300 8-channel
• P20 8-channel

Reagents

• 10x 3' medium throughput v3.1

Optional

• Pipette cam (log liquid transfer errors!)³

Deck Setup

opentrons	deck	setup
10: thermocycler	11: P300 tips	trash
7: thermocycler	8: P300 tips	9: P20 tips
4: mag module	5: P300 tips	6: P20 tips
1: temp module	2: P300 tips	3: 12-well trough

Economic Efficiency

platform	hardware cost	servicing contract	throughput
10x Chromium Connect	$260k	yes	24-samples/day
Opentrons	$26k	no	32-samples/day

Footnotes

1. Running 8 robots for a year, a company can reasonably expect to spend $70 million in reagents and sequence 300M-1.4B single cells. ↩

2. Skipping quantification step for samples with consistent cDNA recovery allows completion of step 2 & 3 uninterrupted. Works great most of the time! Multiplexing always requires P300 tip reload. Tip usage can be further optimized ↩

3. I want to control the pipette with camera input. So much closed-loop precision at our fingertips! ↩