Collection-slice prototype

[johnkerl]

This is a Python-only implementation of SOMA-collection queries like this:

• Dimension-slicing queries:
  • obs_id = b"AGATCTCTATCACG-1"
  • var_id = b"RP11-178G16.4"
  • obs_id = b"HIP002.1512" and var_id = b"RP11-178G16.4"
• Attribute-filter queries:
  • cell_type == "B cell" and feature_name == "MT-CO3"

Such queries will have an optimized implementation in C++ in this repo (cc @gspowley and @ihnorton).

Meanwhile:

• A Python-only implementation gives a lower bound for performance measurement.
• It also serves as a guide for the C++ implementation, so that results may be compared side-by-side.
• Methods implemented here for implementing the prototype, and for doing the kinds of data-exploration to find the examples values of obs_id, var_id, cell_type, and feature_name as shown above, are themselves dogfooding for the SOMA/SOCO API: methods useful for exploring/querying data during prototype development are some of those likely to be useful for users exploring/querying data as well.

[johnkerl]

This is now folded into the greater narrative at #113.