Compare similar packages

[seaaan]

COPASutils
platetools

[jshoyer]

I would be interested in comparison of your package to cellHTS2 -- http://bioconductor.org/packages/cellHTS2

The readPlateList function lets one specify an importFun, allowing the user provide a function that works with the format from their specific plate reader.

[seaaan]

Hi @jshoyer, thanks for your interest. With the caveat that I have scant experience with that package, I see a few differences from plater:

1. You specify the role of each well with a table mapping well ID to role, e.g.,

Well Content
A01 Control
A02 Sample

You may only be able to specify one column for each well?

2. It creates a cellHTS object, rather than a data frame.

3. It is designed with a specific type of experiment (cell-based drug screening) in mind.

The way these aspects work in plater are:

1. You specify well data based on position in the file, so that the file ends up resembling the physical layout of the plate. This can be helpful because you can look at the file as you looked at the plate. On the other hand, the way cellHTS does it might be more intuitive for some people. I find it easier to remember where a well was on the plate than what its well ID was. Additionally, plater can handle an arbitrary number of columns for each well where it looks like cellHTS only has one column (not totally sure though).

2. plater functions return data frames and are designed to work with the tidyverse packages. cellHTS defines its own type of object, which will be familiar to users of other Bioconductor packages.

3. plater provides functions designed to work with any kind of experiment done in a plate and any kind of data. cellHTS seems more focused on a specific type of experiment. I'm sure you could use it more generally, though. cellHTS has the advantage that, for the type of experiment it envisions, it provides a whole suite of other functionality that automates the analysis. plater is much more limited to focusing on data storage, input, and transformation.

Does that help? I can look more into the details of cellHTS if you'd like a more thorough comparison.

[jshoyer]

Thanks---I understand the goals of your package better now.
I should have mentioned that I have run cellsHTS2 before, with my own yeast-based assay data. It is reasonably flexible---I think one could use it with any type of plate reader data, if one wrote an appropriate importFun. One can use multiple channels/columns, as described in the 'multi-channel assay' vignette.

Personally I prefer to specify my plate layout mappings in columnar format (by dragging formulas down in Excel or using the R rep function), but I do also check them in a matrix format (matching the physical plate), so I can see why you prefer that.

I never wrapped my head around the Bioconductor S4 object stuff (phenoData and assayData slots etc.), so I agree that your concept of going straight to a data frame may be more approachable for many users.

[seaaan]

Thanks for the feedback! I feel you on the Bioconductor S4 objects. The idea is appealing to me of having everything wrapped up in a nice container, but I always end up feeling like there's something hidden in the object that I don't remember and that I'm always trying to look in under some curtains to find out what's inside. In that sense a data frame is nice because it's all out in the open.

…

On Wed, Mar 29, 2017 at 2:03 PM, Steen Hoyer ***@***.***> wrote: Thanks---I understand the goals of your package better now. I should have mentioned that I have run cellsHTS2 before, with my own yeast-based assay data. It is reasonably flexible---I think one could use it with any type of plate reader data, if one wrote an appropriate importFun. One can use multiple channels/columns, as described in the 'multi-channel assay' vignette. Personally I prefer to specify my plate layout mappings in columnar format (by dragging formulas down in Excel or using the R rep function), but I do also check them in a matrix format (matching the physical plate), so I can see why you prefer that. I never wrapped my head around the Bioconductor S4 object stuff (phenoData and assayData slots etc.), so I agree that your concept of going straight to a data frame may be more approachable for many users. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#15 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AKDNdWQ2fiEzl9Cmrk9_xzVGj0rp8i3Cks5rqscwgaJpZM4KF_um> .