Methods help, "has this been done before?" and thanks

Hello,

I know that I’ve posted quite frequently in the past few months and and wanted to relay the actual project I’ve been working on (or the main project of several). In the Gregory Rogers lab here at the University of Arizona, condensin is being studied for its role in, you guessed it, condensation of DNA in D. Melanogaster cells. Through manipulation of the various sub-units of the protein as well as blocking of a protein-degrading complex a phenotype has been observed where the DNA turns into these hyper-condensed “gumballs”. I’ve been using the texture analysis of CP to quantify this phenotype and show it to be significantly different than the controls. What I’m wondering may be obvious: has this been done before?

Also, I’ve been tasked with writing a methods section on how the image analysis was done. I’m not sure how much detail to include. In the few papers I’ve skimmed that cited CP all I’ve seen is a few sentences mentioning something like “Image analysis was done using Cell Profiler 2.0” and nothing about pipelines, haralick features or CP Analyst. I want to include a more extensive methods section because CP and CP Analyst have been such great tool for the project. Are there any papers you’ve seen (or written) that have more extensive methods sections? A bit of a novice question: would helping us with the methods section be counted as collaboration or is a citation sufficient?

Lastly, thanks for such a great tool. I’ll be entering the genetics program here at University of Arizona and already know I want to do something in image analysis / computational biology. Your software (and help on these forums) have definitely influenced my research goals.

Hi Scott,

I’m not sure about this one, but I’ll defer to Anne on this one, since she has worked extensively with Drosophila in the past.

I believe a citation will be sufficient. We have a listing of our preferred CP/CPA citiations here, and a listing of papers that cite us here; you can look at the latter to see how others describe our software in their methods.

That’s heartening to hear! Thank you for your diligence and your comments!

Regards,
-Mark

Nice to hear from you, Scott!

I’m pasting below what I think is a good methods section in cases where you must be brief but still convey the basics, adapted from a paper to come out shortly in Cell from the Sabatini lab. This is a very brief mention, just giving the basic gist of the important steps in the pipeline. Depending on the audience of the paper, you would want to define any measurement you use in layman’s terms and perhaps cite a reference for the measurement (see the help for the CP module for references for particular measurements). The pipeline should then be included as supplementary data, and/or can be posted on your lab website, or at the CellProfiler website (cellprofiler.org/published_pipelines.shtml). Part of our motivation for the design of the CellProfiler project was to enable sharing of small pipeline files, to aid in reproducible research and to save researchers’ time trying to reconstruct prior work.

It is not uncommon for researchers to feel the time spent and the intellectual input into the image analysis for the paper is more significant. In such cases, you might devote a paragraph or two in the main text of the paper or just lengthen the description in the methods section, describing the main modules and important settings in narrative form but skipping such details as the numerical settings. Sometimes entire articles are written solely about developing an approach that works for a single assay! In such cases there is usually significant validation for each step in a pipeline, or more focus on the wet work that went into creating the assay.

Regarding your question about studying chromatin structure by image-based texture analysis, I’m not aware of papers that specifically have studied condensin in this way; however, I finished my PhD in the field of chromatin structure almost 10 years ago, so I would not take this assessment as definitive! Texture measurement from images is more common in histology/pathology than in fluorescence microscopy/basic biology research. Search in Pubmed for “texture chromatin”, perhaps qualified by the phrase “image analysis” and you will see some examples.

Best wishes in your future research!
Anne

Nuclear Eccentricity and Lipin 1 Nuclear-Cytoplasmic Proportion Measurements

Nuclear eccentricity was measured with CellProfiler[citation: Genome Biology paper
PMID: 17076895] (www.cellprofiler.org), version r10997 (pipeline provided as
supplementary data). Image analysis was performed using
10x images with the eccentricity measurement being made on nuclei identified by
lamin A immunostaining. FLAG-lipin 1 nuclear proportion was quantified by
CellProfiler as follows: after illumination correction, the nuclei were automatically
identified using the DAPI staining; the cytoplasmic compartment was defined by
expanding the edges of the nuclei 20 pixels in every direction; nuclear proportion
was then measured as nuclear intensity/(nuclear+cytoplasmic intensity).