How Ochre Bio automated a manual image processing workflow with Nextflow

About Ochre Bio

The Ochre Bio team comprises a diverse, interdisciplinary group of scientific experts working on chronic liver diseases. They faced a challenge in moving some of their work forward because of a dependency on a manual image-processing workflow, compounded by working across different locations and time zones.

The Challenge

One of Ochre Bio’s Taipei-based computational biologists would run a workflow to support a team of wet lab scientists based in New York City (NYC). This involved image pre-processing in Python to divide large images into smaller ones, characterizing image tiles using the open-source CellProfiler software, then aggregating results and generating plots for visualization with R.

/low-visibility.png?length=720&name=low-visibility.png)
Low visibility

As with most organizations, infrastructure standardization across different teams posed challenges.

/bottleneck.png?length=720&name=bottleneck.png)
Bottleneck

Work could only be done by a single computational biologist, working in a different location.

/time-constraint.png?length=720&name=time-constraint.png)
Time constraints

A multi-step process across different time zones meant increased waiting times.

Created separate Compute Capsules for image preprocessing, running CellProfiler, and results visualization

Created a Pipeline to run multiple instances of the CellProfiler Capsule in parallel on AWS Batch

Created a parameterized no-code version of the pipeline to share with NYC-based colleagues who could use it with no additional support

Ochre Bio automated their manual workflow by moving it into Code Ocean

The Results

Productivity

Automating this process has freed up Ochre Bio’s computational biologists to work on other tasks for the org

Performance

Demanding computational steps are now automated and parallelized, making the overall workflow more robust and significantly faster

Access

Wet lab scientists are now enabled to run computational pipelines with minimal guidance from the dry lab, democratizing access to tools

Visibility

The entire workflow is now visible to the rest of the Ochre Bio team, and isn't locked into any one particular platform (can always be exported)

“Code Ocean sped up our internal image pre-processing computational workflow by at least 10x, improving collaboration and productivity between our global teams across time zones and removing the need for complicated and painful cloud computing infrastructure set up.”

Dimitris Polychronopoulos

Director of In Silico Biology, Ochre Bio

Dimitris Polychronopoulos

Director of In Silico Biology, Ochre Bio

Key features used

Capsules

Provision, develop, run, and share. Always reproducible. Learn more

Pipelines

Chain Capsules together into parallelized workflows. Learn more

No-code Apps

Deploy Capsules & Pipelines as parameterized apps Learn more

Data

Attach EFS-cached data already in your S3 bucket. Learn more