spacetx/sptx-format

Image data format specification for spaceTx

sptx-format's Issues

Ensure package installation installs non-python files

use, e.g., MANIFEST.IN with include_package_data

see https://github.com/freeman-lab/showit

hybridization_images -> primary_images

Change hybridization images to primary images. Propagate these changes through the other repositories, potentially through creating constants in this package that are importable.

determine how to version breaking changes

need a plan that states how we will increment versioning.

Investigate using linked schema

@hewgreen suggested that HCA metadata has outlawed the type of implicit linking we have in our specification.

Example experiment.json

{
  hybridization_images: "hybridization.json",
  ..., 
}

In existing specification, we link a file which is implicitly of a particular type. Here it can be either an fov_manifest.json or a field_of_view.json.

HCA metadata suggests making it an explicit link, wherein it should contain the metadata structure. Investigate the pros/cons of this for our use cases.

move primary_images validation to slicedimage

move primary_images validation to slicedimage

FieldOfView tile_shape has ambiguous (x, y) ordering

The schema claims that it is ordered (x, y)
https://github.com/spacetx/sptx-format/blob/master/schema/field_of_view/field_of_view.json#L91

Internally, we don't validate on this field yet, we simply slurp in the data and make sure that the sizes fit. However, we might want to provide information on a TileSet (before it's loaded into memory).

We should replace this with an explicit map of {x: size, y: size}, instead of a list to prevent errors.

Move Indices to sptx-format

Starfish constants should live here so that they are only updated in one repo.

Implementation possibilities include:

• having importable python objects
• using them to fill Jinja template json schema that are rendered by CI

Write spacetx validator

We should have a validator for spaceTx format so that we can:

1. verify tiles are valid early, providing faster feedback to users
2. enable collaborators to more easily understand and consume data in our specification.

Specification needs textual description in form of README.

We should lift some of the prose from the below document and update it into a README that can be provided to spaceTx collaborators.

https://docs.google.com/document/d/12V-BQF-wh1GDBXwUH2hQrPhw6n0oQqENPK-YHGv59w4/edit#

Codebook needs a channel map

The codebook should associate information about the channel ID with the channel indices to enable user-driven QC.

Codebook needs a version field

Codebook is currently unversioned.