PHOTONMOSAIC: A FLEXIBLE, MODULAR AND INTEROPERABLE SOFTWARE FRAMEWORK FOR MULTIPHOTON IMAGING DATA

In vivo multiphoton calcium and voltage imaging allows activity to be measured across specific neural populations, and continues to evolve through experimental advantages such as multi-plane and multi-region acquisition. Several high-quality software tools (e.g. Suite2p, CaImAn, …) to process, analyse and visualise this data exist. Yet the growing scale and complexity of these datasets expose limitations in today’s fragmented software ecosystem. It remains therefore difficult to combine and compare individual steps of established pipelines, or replace them with new implementations tailored to specific cutting-edge experiments. Here, we introduce our work on PhotonMosaic, a community-driven project which aims to address these shortcomings.

Firstly, we are developing a standardised Python API for widely-used steps (such as motion correction and cell detection). The Python API allows users to run various algorithms via their own Python scripts or integrate it in their custom packages. We aim for a modular and extensible design, and borrow concepts from the popular electrophysiology analysis software SpikeInterface.

Secondly, we are developing pipeline tools to automate, parallelise and standardise the execution of these steps, also on high-performance computing systems.

The PhotonMosaic project is in its early stages. We hope that with the community’s support, PhotonMosaic can become the “glue” that allows multiphoton imaging software tools to be combined and compared more easily, and built upon in a reproducible way. We encourage everyone with an interest in PhotonMosaic to give feedback and contribute to these efforts, by engaging on the online forum, the Zulip developer chat and on Github.