A general comment about the Example data sets



I’ve been using the Example data sets as a tutorial to learn CellProfiler. Along with the recent publications, it’s been a useful way to figure out exactly how everything works and how the various parts fit together. I especially enjoyed working through the “Speckles” example, which, in addition to being very close to an application we’ll use in our lab, covers a lot of territory and is fun to watch.

Since I’ve gone from naive n00b to designing my own pipelines, I thought I’d take the opportunity to discuss my new user experience as I worked through the examples.

There were two negative aspects of this experience that I wanted to share (and eventually express as positive suggestions):

  1. Several of the examples have pipelines that refer to old versions of specific modules, and require the user to figure out how to translate the old parameters into the new parameters. In these cases, the resulting confusion for the new user results in quite a bit of frustration. How should I know how to set the parameters of “Crop” on a pipeline whose ultimate purpose I don’t know? Which brings me to …

  2. Since there’s no commentary or explanation of the choice of modules, the new user confronted with pipelines for the first time has to figure out what each one does, how the decision was made to include each module, and how the parameters were set for each one.

That’s the criticism. Here are my suggestions:

  1. Update the Examples so that they correspond to the newest version of the modules, and keep them current, so that they stay the most didactically useful. In particular:

ExamplesHumanImages: old "Crop"
ExamplesFlyImages: old "Crop"
ExamplesSBSImages: old “CalculateStatistics” and “CalculateRatios”

  1. Accompany each example with a concise description of the purpose of the pipeline and a few words about how the pipeline was designed. There are many cases where I would have learned faster if I knew why the pipeline was the way it was, instead of some other way.

  2. Either by adding a line item to the description of this sub-forum or by creating a new category, I think a useful forum category would be (e.g.) “Pimp My PIPE” – a forum devoted to people who are designing pipelines and want (constructive) criticism from others about their pipeline design. (“Am I doing this right? Could I be doing it better?”) In addition to being useful for individuals, this type of discussion might be extremely useful to new users, as it would give them a chance to see the development of their peers’ thought processes.

Long-winded, I know, but I hope it’s helpful. I’m sure I’ll be posting again soon as I design more ambitious image processing projects.

To the developers: thanks for maintaining the great site, and for all the work you put into CellProfiler.



I think you have made some good points. The pipelines should definitely be kept up to date and have only slipped through the cracks with the last two releases because we are currently very busy! The CellProfiler project will be moving with Anne Carpenter to the Broad Institute. When this happens, there will be more support for users such as yourself to get help with their pipelines. However, I will try to post updated pipelines as soon as I can!