HADES utility package for standards IO operations and create

[azimov]

Currently, it is up to individual package maintainers to decide on how to use different methods within R to do basic operations like serialization and deserialization of JSON objects, reading and writing csv files, indexing covariates and cohort ids for bulk operations etc. What would be useful is to agree on a set of best practices for these operations and implement these in a standard R package that could be used across the HADES suite.

To start the list for initial utilities/standards:

• A toJSON and fromJSON standard that exports R data.frames in a consistent manner and consistently unboxes variables. For example, jsonlite and RJSONIO are used throughout packages and, in some cases, both packages are used. A common function should be used.

• Read/Write csv - it is common to see different R packages to do this leading to inconsistent csv formats that often require bespoke methods to reformat data files

• A consistent way of defining id's for cohorts and covariates based on concept sets. For example, there are many standards for this (e.g. concept_id * 1000 + <some small int>) to denote cohorts of different types. It would be useful to define ways for creating ids and tables that allow composite keys with mappings.

Please comment with any further requirements we may have for such a package.

[anthonysena]

[HADES Hack-a-thon]: @azimov is going to work on this item.