identifiers-and-datasets

library(coconat)

datasets

coconat needs to be able to handle a range of datasets. For each dataset we record both a long name and a short name (2 characters [a-z0-9]) that we use to refer to them.

You can tell coconat about a new dataset by doing:

register_dataset('flywire', shortname = 'fw', species = 'Drosophila melanogaster', sex = 'F')

identifiers and keys

Each dataset will have unique identifiers for the neurons that it contains. All connectomics data sources that we currently work with have identifiers within the 64 bit integer range, but there is no reason that they cannot contain additional characters. To ensure interoperability, we will

1. default to handling all identifiers as character vectors (strings) even if the native form in a package providing access to a dataset is some king of numeric.
2. introduce the notion of a key in the form "<dataset>:<id>"

The second point is vital since the same numeric identifier can be used in different datasets to refer to different neurons.

We provide a number of functions to assist with this. At the most basic level id2char()will convert numeric ids to a character representation taking care of issues like the fact that as.character(1e5)=1e+05 which can cause trouble.

id2char(1000)
#> [1] "1000"
id2char(1e5)
#> [1] "100000"
id2char(NA)
#> [1] NA
id2char(bit64::as.integer64('12345678901234'))
#> [1] "12345678901234"