ratschlab-common 0.1.0

Small library of common functionalities used in various projects in the ratschlab

Small library of common code used in various projects in the ratschlab.

• Free software: MIT license

Features

• Writing parquet and HDF5 files with sensible defaults.

• Support for working with ‘chunkfiles’, i.e. splitting up a large dataset in smaller chunks which can be processed independently:

  • Repartition records (i.e. increase or decrease number of chunkfiles) while keeping data belonging together in the same file (e.g. data with the same patient id associated)

  • simple indexing for looking up in which chunk to find data belonging e.g. to a patient

• bigmatrix: support for creating and reading large matrices stored in HDF5 having additional metadata on the axes in form of data frames.