A few notes on how data is organized in episb-provider / Elasticsearch

Several key concepts and data structures we implement are: Segment (region), Segmentation (list of regions), Annotation. There are others (such as Experiment, Study and Author) but we will focus on the first three below.

Segment (region)

A region is characterized by a Segment ID, the chromosome it is on, the start and end of the portion of the chromosome.

The JSON "view" of an example region structure is as follows:

   "segID": "testsegmentation::6299bb3d-9d07-4b45-94ae-79c3f96ef6e9",
   "segChr": "1",
   "segStart": 12137,
   "segEnd": 14537

The segID field above is a combination of the segmentation this segment belongs to (portion before "::") and a UUID randomly assigned to the segment at the time it was entered into the database (or in other words, "created"). We can then easily deduce the segmentation this segment belongs to from the segment ID.

Segments live in the "segments" index in Elastic.


is simply a list of Segments. It is committed in Elastic like so:

    "segmentationName": "testsegmentation",
    "segmentList": [

Segmentations live in the "segmentations" index in Elastic.


One of the main concepts of this project is to separate annotations from the segment information. In other words, we can try and provide more "universal" segmentations (lists of segments) that hopefully, experiments can "subscribe to" and use. An annotation is then a value (as arrived at by an experiment - could be an integer, a categorical value or a float, for example) and a segment ID from a particular segmentation.

For example,

    "segmentID": "testsegmentation::8b74bed5-c001-4cd5-aa3f-aa07e2bdfdc0",
    "annValue": "0",
    "experiment": {
        "experimentName": "testexperiment",
        "experimentProtocol": "",
        "experimentCellType": "",
        "experimentSpecies": "",
        "experimentTissue": "",
        "experimentAntibody": "",
        "experimentTreatment": "",
        "experimentDescription": "Loaded from preformatted file"
    "study": {
        "studyAuthor": {
            "familyName": "episb",
            "givenName": "default",
            "email": "[email protected]"
        "studyManuscript": "",
        "studyDescription": "",
        "studyDate": ""

Annotations live in the "annotations" index in Elastic.

Provider interface

The purpose of this JSON structure is to describe a provider. Each data provider serves this structure at the /provider-interface URL path. The contents of the structure are filled in from episb-provider/episb-provider/src/main/resources/application.conf - a HOCON file that can (and should) be manually edited (example).

Experiment/segmentation pairing structure

This structure tells a user (an episb-hub, for example), what segmentations the episb-provider is able to serve, but also all the experiments subscribing to these segmentations (internally this is implemented as a list of JSON objects - an example of which is below). The document describes the name of the segmentation, the name of the experiment subscribing to the segmentation, some auxiliary information and tells the hub how to get the annotation value and the range this value can take.

For example,

    "providerName": "episb-provider",
    "providerDescription": "sample segmentation provider",
    "segmentationName": "testsegmentation",
    "experimentName": "testexperiment",
    "cellType": "sample cell type",
    "description": "sample experiment description",
    "annotationKey": "value",
    "annotationRangeStart": "0",
    "annotationRangeEnd": "1"

The list of these documents lives in the "interfaces" index in Elastic search. API calls exist to get this list or search it for a particular segmentation, for example.