swarchal / columbusimporter Goto Github PK

Should be able to:

• work on linux
• pass parameters by command line
• pass parameters by yaml

Currently it's using a lot of memory after the csv file has been saved. This memory is not cleared even if the program is ran again to create a csv on a new dataset.

I think should should have been caught by the GC but there's a reference to the data hanging around somewhere which might need to be explicitly deleted.

e.g here

As they're going to be removed when compiled with -d:release.

Causing too many problems as it's not included in the binary.

Since the regex is used on fixed-format filenames I can just hard-code the parsing functions.

Currently it saves a csv file when the incucyte parser is used on CV8000 paths.

This is because the metadata parsing is simply slicing strings at fixed positions.

There should be some sanity checks for the filepaths and an error should be returned if the paths don't look like what the parser is expecting.

If we're using the output from FNet then the filename will contain a filename:
EXPERIMENT_WELL_T001F001L01A01Z01C01.fnet_name.tiff
which causes the parser to fail when extracting the metadata.

Maybe a more robust approach would be to only take the filename before the first ..

The measurements file is used to get the image resolution, but if we're faking images acquired by the CV8000 then this will not be present.

If it's not present then it would be useful to insert a fake .mrf file for the correct objective to get the x-y image resolution.