Hi @Konrad Rokicki Thanks so much for creating this channel and starting the discussion around tools and pipelines from image analysis in nextflow and nf-core! My name is Florian, I am a Postdoc with Denis Schapiro at University Hospital Heidelberg (https://www.schapirolab.com/). As you might have seen from the recent hackathon, we are working a nextflow pipeline called MCMICRO (https://mcmicro.org/) and have been putting in some work during the hackathon to add tools from MCMICRO as modules to nf-core. We are also always working with new imaging technologies and methods and love contributing pipelines for these. I am currently working a lot with multiplexed FISH methods, of which many new companies are coming out at the moment, all of which deliver different data types at different levels of processing (some provide spot images, still to be processed, others provide ready to assign spot tables. Some provide segmentation approaches and masks, others don’t). I was thinking about whether it would make sense to suggest creating an nf-core pipeline for general analysis of multiplexed FISH based assays using nextflow. Different workflows could be geared to the different technologies to help end-users have some guidance on how they can get started. What do you think about this? Do you think it’s different enough from your multifish pipeline to warrant its own pipeline (since in most cases, users work with 2d data, not that often with expansion 3d data)? Very much forward to discussing with you in the future here 🙂 !

Hi @Florian Wuennemann, great to meet you and thanks a lot for sharing all this! I think this is exactly the type of thing we should be discussing here to find potential points of alignment. I’d be very interested to learn more about the methods and software packages you are using for multiplexed FISH. EASI-FISH is just one of many multiplexed FISH methods out there, and it seems there are more being developed all the time. Some methods might require very specialized computational pipelines (e.g. like nf-core/spatialtranscriptomics is intended for 10x genomics), but I’m sure it makes sense to group many of the FISH methods together into a generalized pipeline as you suggest.

Very much agree

Hi @Konrad Rokicki, sorry for the belated reply! I absolutely agree with everything you wrote above and think that there are a host of different applications that lend themselves to the pipelining and modular infrastructure of nf-core and nextflow. I also agree that it is very challenging to know in advance, what future methods will require in terms of processing, file formats etc. I see a number of different technologies/methods that are currently in use and development, where there is some overlap in processing and some where completely different inputs, outputs and processing is needed. Here are some technologies that have been mentioned in the channel, that I think require their own pipelines: • lightsheetmicroscopy -> nf-lightsheetrecon @Konrad Rokicki + pipeline idea from @Luis Kuhn @Carolin Schwitalla • FISH based assays -> EASI-FISH @Konrad Rokicki (geared specifically for that setup, lots of processing done), I am working on a pipeline for Molecular Cartography data from Resolve Bioscience (less processing of raw data needed, as spot counts are provided, but still need a bit of image processing, segmentation and assigning spots to masks) • Highly-multiplexed antibody based assays (Cycif, Akoya, Lunaphore, etc.) -> MCMICRO + MCMICRO translation to nf-core @Adam Taylor, @Jeremy Muhlich among others and our lab in Heidelberg (Schapiro lab) • A combination of RNa + protein based assays Any other that I forget that are not yet pipelines in nf-core (like VISIUM spatial transcriptomics that exists already) ?

@Florian Wuennemann, that's a great outline of the current efforts and I think there will be many other types of microscopy pipelines in the future, especially if we can pave the way 😁