Debug bioinformatics workflows in the cloud from your local development environment
iteratively debug remote code // dispatch unit tests to the cloud // interactively debug production environments from your laptop
We have written previously about the LatchBio SDK, a python bioinformatics library deployed by dozens of biotech companies to manage cloud infrastructure and accelerate science.
After working closely with a few dozen of our customers and piloting software tooling, we realized the largest bottleneck to developing bioinformatics workflows was establishing a tight feedback loop with a production cloud environment for fast iteration and debugging.
Today, we are augmenting the toolkit by allowing developers to easily debug biological code in the cloud from the comfort of a local development environment.
Why we need DevTools for biology
Biotech companies run many experiments that spit out a lot of unintelligible data. The class of programs that process these uninterpretable files and extract latent biological insight are called bioinformatics workflows. They are difficult to develop because they need the computing resources provided by large cloud machines and gigabytes of input data to properly reproduce an error or test behavior.
The experimentation engine of a lab or company is completely dependent on the efficient development and use of these programs, where inefficiencies in engineering can delay the arrival of drugs to the clinic.
Debug biological code in the cloud from your laptop
After working closely with a few dozen of our customers and piloting software tooling across different types of biology in industry and academia, it was clear that the largest bottleneck to developing bioinformatics workflows was establishing a tight feedback loop with a production cloud environment for fast iteration and debugging.
Such a feedback loop gives access to the following resources that are particularly problematic in bioinformatics:
bespoke libraries / dependencies only available in a cloud environment
access to large files needed to reproduce errors
enough cloud resources - cores, memory, GPUs - needed to even run the program
We added utilities to the LatchBio SDK to quickly gain access to these resources for rapid debugging.
Get an interactive shell
When programs misbehave, developers can reach for a shell to freely explore the environment and filesystem.
Interactively debug production cloud environments
Developers can write and edit python scripts locally, run them in a cloud environment, inspect output and iterate with error logs.
Dispatch local unit tests to the cloud
Unit tests can be written as local scripts and executed directly in a production environment to verify workflow behavior is consistent across changes.
We’re stoked to release these tools to the community to see how people build biologicals software in new ways. The code is available here. Documentation here.
At LatchBio, we believe strongly that mathematics and computer science will transform the life sciences. Our team is obsessed using deep technical knowledge to build a lasting software platform that will dramatically accelerate the rate at which we can engineer biology. If you are an earnest engineer, consider joining the team.