Programming the wet lab

For most of its history, computational biology stopped at the edge of the bench. That boundary is dissolving: the wet lab is becoming programmable, and reproducibility is following.

Robots you write in Python

Opentrons pioneered affordable liquid handlers you control with Python. A protocol is just code, which means it can be versioned, reviewed and shared like any other software:

1def run(ctx):
2    plate = ctx.load_labware("corning_96_wellplate_360ul_flat", "D1")
3    tips = ctx.load_labware("opentrons_flex_96_tiprack_200ul", "C1")
4    pip = ctx.load_instrument("flex_1channel_1000", "left", tip_racks=[tips])
5    pip.transfer(100, plate["A1"], plate["B1"])
Copy

The Flex Python API keeps growing — recent versions added motor and gripper control, concurrent module actions and finer control over transfers.

One interface for many machines

If you run a mixed fleet, PyLabRobot gives you a single, hardware-agnostic Python API across Hamilton, Tecan and Opentrons devices — plus plate readers and pumps. You can simulate a protocol before it ever touches real hardware, which saves reagents and prevents expensive mistakes.

Closing the loop with the notebook

Automation is only reproducible if the records are too. Modern electronic lab notebooks — commercial platforms like Benchling or self-hosted open-source options like eLabFTW — capture protocols, samples and results with APIs that let your pipelines read and write experimental metadata directly.

The payoff

When a protocol is code, an experiment becomes a pull request: diffable, reviewable and re-runnable. That is the same shift software engineering went through decades ago, finally arriving in biology.