Failed to import vigra library

Newbie here: I just downloaded regular user CellProfiler 2.1.1 on my Windows 7 system. The small black background message windows says:" vigra import: failed to import the vigra library ".
I looked at the suggested Webpage, and I am not enough of a programmer to figure out what I need too do to fix the failure. (Although I did learn that if you Google vigra and don’t spell properly, you get some interesting results.)

Hi,

That’s odd. Are you using ClassifyPixels? Or do you mean you get that warning when you first launch CellProfiler?

I believe that ClassifyPixels is the only module that uses vigra (and yes, that library was not optimally named!). So if you are not using ilastik and ClassifyPixels, this warning can be ignored. Though it is odd to get it at all if you are just using the compiled “standard” version of CellProfiler, so if you want to clean the slate first, you can try this: Close down CellProfiler and clear CellProfiler’s Windows registry entry, as in this post:
cellprofiler.org/forum/viewt … 3c2#p11494

Hope that helps,
David

Thanks for helping a ‘hand wringing’ new user. Yes, I get the warning on launch of the compiled “standard version” – that was my confusion – why was it asking for references to what the Webpage says is essentially an include file ?
Even though I don’t use (yet) the ClassifyPixels , I did delete the registry entry. I will just ignore the warnings and get on with it.

I tried the “Load an example pipeline” under the “See the a pipeline in action” heading at the beginning of the Welcome Page, and it works fine, and is the basis for teaching myself Cell Profiler. It is a fantastic program, and I’m sorry that I learned of it only recently –

I don’t have Plate Reader data, but do have a lot of *.LSM and *.ZVI images which have soaked up countless hours of mousing, pointing and clicking for ‘by hand’ image analysis . With Cell Profiler, I hope to move forward with automation, and never look back.

To that end, I have one further question – I looked at the Section “Build your own pipeline” on the Welcome page, and perused the " 1. Start Download a pipeline template…" Under the “More Advanced Pipelines”, I found Human cytoplasm-nucleus translocation assay (SBS Bioimage) example. This is EXACTLY what I need too do in my work (and avoid carpal tunnel), so I want to study this example top to bottom. But, when I load the example, Cell Profiler balks at the MetaData module level. Even though I can browse for the CSV file, and find it, the module still says it can’t find it !

I did download and read the Users Manual (RTFM !) and it mentions some incompatibilities with pipelines generated with versions earlier than 2.1 – is this the issue with this pipeline ?

Hmm, you seem to be doing it right. I just tried this and it works for me on my Mac. I did this:
(1) Drag ExampleSBS.cppipe to CP 2.1.1 pipeline window. It complains about being from an older version, but it should work ok.
(2) Drag in all the .tif files to the Images module.
(3) In Metadata, click the folder icon and locate the “1049_Metadata.csv” file. It takes a few seconds, but the red warning text goes away once the csv is found and validated. ** Make sure the CSV is not open in Excel or any other software that might be locking the file (though I doubt this would be a problem).
(4) Click the Update button in Metadata and inspect the table it produces.

I didn’t go farther, but Is this what you are doing? If so, what OS are you using?

David

Thanks for taking the time to investigate. In fact, there is no problem with my CSV Metafile loading. If I could delete the post, or add an emoticon that connotes "slinking away’ I would do either of those. If you want add a FAQ for "Incredible Rookie Mistakes, here is what happened: Because of my screen size and resolution, the Metadata module loads on my screen with the Update section exactly covering up the last section on loading the CSV file. I never saw it, and since the manual does not have a complete screen capture, I didn’t realize it existed, and tried to modify the earlier extraction profiles – D’oh ! Finally, noticed the scroll bar,scrolled down and there was the section ! If want to send me out to get you a Left Handed Compiler, or a Sky Hook Debugger, I’m up for it.

Ha - no problem. If you have suggestions on how we can make it more user-friendly, let us know. The scroll bar in the settings panel has been missed by more than one person, I know that for sure! And if the manual could be improved with a full screenshot, etc let us know.

Cheers,
David

I am continuing my adventures with loading Metadata with a .CSV file. Although I can load and run the CP example files, when I try to load my own CSV, (with my own images) I get a puzzling result with my Doses and Controls columns as generated by the update button. As you read down the columns, these cells start with “Not Available” at the beginning of the image numbers, then resume with accurately matched { ImageNumber, Doses, Controls} information, and then ends with the Doses and Controls as “Not Available”.

I have checked the .CSV file with a non-formatting word processor (Notepad) and no strange characters appear anywhere in the file.
I generated the file by creating an Excel spreadsheet with the appropriate information, then saved as CSV file.

Would you mind posting the csv? We’ll a take a look.
Also, please post the pipeline too, since there are different ways to load metadata (e.g. LoadData module, Metadata module).

Oh, and did you choose “Windows Comma Separated Values” when you exported from Excel? On Macs at least, you need to be careful how you export, though I can’t remember what the options are for Windows.

David

MetaData_XLS.csv (364 Bytes)YAPCRT1.cpproj (533 KB)Thanks for taking a look. Here they are

Not 100% certain what the issue is, but I would not match on ImageNumber. In fact, remove the ImageNumber column from your CSV. And match only with metadata that you provide and that CellProfiler wouldn’t know about otherwise. (ImageNumber is integer data type, not “Text” which also might have been an issue.) Further, ImageNumber is assigned by CellProfiler in a way that might not correspond to your image list order, especially since CP now has multicore-processing and the order is even murkier, but the user shouldn’t have to over-think this. Match the metadata that uniquely specify a physical object, like a well, or a site within a well. So in your case, “Well” should do it here. But note that if you run multiple plates, or multiple sites within a well, you’ll need to specify/match on those other metadata (i.e. Plate, Well, and Site).

See if that helps.
David

That solved the problem !
Thanks for help and the advice about setting the matching.
I’m enjoying CellProfiler very much, and have a large amount of images I want to Profile – and now I can !

Sorry for the double post, but I wanted to ask about the CellProfiler generated Dose-Response Plots – in addition to the fitted curves, why is it that expected data values rather than the observed values are plotted ?

To “close the loop” on the topic I started, I have just loaded 64-bit CellProfiler into a 64-bit Windows 7 system, and I do not get the "Failed to import vigra library " message. My previous load was a 32-bit CellProfiler into a 32-bit Windows 7 system.

Ah, ok, our 32-bit version is being phased out soon (and this error may have been one reason why!).

As for why are the “expected data values rather than the observed values are plotted”: I’m not sure what you mean – can you provide a screenshot of the CalculateStatistics output (which is where this output is coming from I assume)?

Thanks
David

Sure, here is a plot. I have verified (by looking at averaged raw data) that the plotted line is a good fit to the datapoints. However, what is plotted as data points (the green dots) fit the curve exactly, that is, the line goes directly through the center of the dots. The actual average points do not fit exactly on the line - some are above, some are below. It seems that it would be more useful to plot the fitted curve, as happens, but then to point plot the actual averaged data, not not points that would be expected (i.e, fit exactly to the curve) at a given dose.DoseResponsePlotDistCytoplasm_Math_Ratio1.pdf (12.9 KB)

I see, thanks. Yes, your suggestion is totally valid. I haven’t looked at these plots for quite some time, but I seem to remember long ago they used to plot the data as they ought to, rather than the points along the fit curve. In any case, I’ll add this to our Issues list.

Thanks,
David

Thanks – and to end this thread on a positive note, I have now tried most of the CellProfiler examples ( I got them all to work as they should) and have taught myself a lot about CellProfiler in the process. I was pleased to see the Wound Healing example, as we have done this type of experiment in the past, and will do so in the future. I look forward to the analysis with a lot less carpal tunnel stress on me. I hope 2015 will bring several CellProfiler citations from our lab.