Save / Load with multiple data inputs / outputs

[kescobo]

trafficstars

I'm frequently dealing with a really weird tabular data structure called a PCL file - basically, the first n rows contain various forms of metadata (first column for these rows is metadata name), then below that it has a sample x feature matrix. So for example, I might have my_data.pcl:

age	12	14	18
gender	m	f	f
sampleID	a	b	c
feature1	0.2	0.4	0.3
feature2	0.3	0.2	0.3
feature3	0.1	0.6	0.7

Since most of the operations occur on the numerical table part, and storing this all in a single dataframe (or whatever) would generally lead to columns with type Any, what I'd like to be able to do is have load/save functions that make/take two iterable tables, that share the sampleID row, eg the two tables would be:

metadatadf:

sampleID	a	b	c
age	12	14	18
gender	m	f	f

featuredf:

sampleID	a	b	c
feature1	0.2	0.4	0.3
feature2	0.3	0.2	0.3
feature3	0.1	0.6	0.7

And I'd like to be able to do something like:

(x, y) = load("my_data.pcl", id_row="sampleID")
metadatadf = DataFrame(x)
featuredf = DataFrame(y)

save("new_table.pcl", metadatadf, featuredf)

Can I get some guidance on whether this is possible / makes sense to use the FileIO framework for this?

[SimonDanisch]

If I'm not mistaken, arbitrary signatures should be supported in FileIO - just make sure that you accept those signates in your IO library! If FileIO misses to pass down the keyword args or additional arguments, please open an issue!

[kescobo]

Great! An orthogonal question that just occurred to me - should I / can I piggyback off of the functions already present in CSVFiles.jl? Since this is a special case of a csv/tsv file type, and I'll probably want to make use of all of the keyword stuff available there?