HCA Ingest Service neo4j graph validator package
HCA Ingest Service Graph Validation Suite
What is this useful for in the scope of the HCA:
- Enables data wranglers to visually analyze the relationships inside a submission to look for inconsistencies.
- Provides an automated graph validator for which to create tests using step 1 and can be run fully containerized.
The suite is divided in two separate, extensible parts:
hydrators enable users to import and populate data into a graph database. The reason not to call them importers is
importis a reserved keyword in Python and
from importers import importeris a bit confusing. :dizzy_face:
actions provide different tools to work with the generated graph. The first and most important is to run a series of tests to validate the constraints Data Wranglers want to impose on submissions. Another action is generating reports and extracting statistics from the graph to send to the submitters. Any other actions can be implemented to extend the suite.
So far, the functionality planned is as follows (WIP items are still not fully implemented):
- Ingest Service Spreadsheet.
- Ingest Service API Submission.
- BioSamples API (WIP).
- Opening an interactive visualizer to query the graph.
- Running tests on the graph.
- Generating reports for the graph (WIP).
The Graph Validator Suite requires docker running in the host machine.
From the git repo
git clone email@example.com:ebi-ait/ingest-graph-validator.git cd ingest-graph-validator pip install .
A Python package has been published in (PyPI)[https://pypi.org/project/ingest-graph-validator].
The installation into a (virtualenv)[https://packaging.python.org/guides/installing-using-pip-and-virtual-environments] is heavily recommended.
pip install ingest-graph-validator
The graph validator is installed on the Wrangler EC2. The Neo4j Browser UI of this instance can be accessed at http://tool.archive.data.humancellatlas.org:7474/browser/.
If you have HCAWrangler user credientials configured in a profile called
hca-util, the username/password stored in AWS Secrets can be retrieved using the command:
$ aws secretsmanager get-secret-value --secret-id ingest/tool/wrangler/ec2/neo4j --region us-east-1 --query SecretString --output text --profile hca-util
To import/load new project into the graph db:
log into the EC2 via SSH
navigate to the installation dir and activate the environment
$ cd /data/tools/ingest-graph-validator/ $ . venv/bin/activate
you don't need to run
ingest-graph-validator initas the server is already running in the background
you will need to run the following command before importing a spreadsheet or ingest submission uuid, where password is determined using the aws secretsmanager command shown above (password includes single quotation marks):
$ export INGEST_GRAPH_VALIDATOR_NEO4J_DB_PASSWORD=[password]
import a spreadsheet
$ ingest-graph-validator hydrate xls <path/to/xls>
import using an ingest submission uuid
$ ingest-graph-validator hydrate ingest <sub_uuid>
run tests (commonly used dir is graph_test_set)
$ ingest-graph-validator action test <path/to/tests_dir>
don't forget to exit the virtual environment when done
Step by step usage for running locally
Ensure Docker is installed and running
Once installed from the Python package, start the backend by opening a terminal and typing:
Keep in mind, first time executing the
initcommand will take longer as it has to pull the Neo4j Docker image from dockerhub.
Import a spreadsheet:
ingest-graph-validator hydrate xls <spreadsheet filename>
Go to http://localhost:7474 in a browser to open the frontend.
Connect to the backend (you do not need to change any fields, leave username/password empty):
You can then start writing cypher queries in the input field on top of the web frontend to visualize the graph. For example:
MATCH p=(n) RETURN p
Will show the entire graph. Keep in mind this will crash the browser on huge datasets.
Note The server backend will continue running in the background, and you only need to open the browser again to continue your work. If you want to shutdown the backend, open a terminal and type:
How to run tests
In order to run the tests in the
graph_test_set directory, you need to run the following commands:
- In a shell, run:
If you want to run using an ingest submission uuid:
ingest-graph-validator hydrate ingest <ingest_submission_uuid>
If you want to run using a spreadsheet
ingest-graph-validator hydrate xls <path/to/spreadsheet>
ingest-graph-validator action test <path_to_tests>
The Graph Validator Suite uses a CLI similar to git. Running a command without specifying anything else will show help for that command. At each level, the commands have different arguments and options. Running any subcommand with
--help with give you more information about it.
The root level commands are:
ingest-graph-validator initstarts the database backend and enables a frontend visualizer to query the database, in
ingest-graph-validator hydrateshows the list of available hydrators.
ingest-graph-validator actionsshows the list of available actions.
ingest-graph-validator shutdownstops the backend.
Useful cypher queries
Show all nodes and relationships
MATCH p = (n) RETURN p
Show all nodes and relationship excluding some
MATCH p = (n) WHERE NOT n:LABEL AND NOT n:LABEL RETURN p
Expands paths from a node
This one will be shown with an example. The example selects the donor CBTM-376C from Meyer's Tissue Stability dataset, and expands the paths to show all biomaterials, processes and files linked to it.
Note: Make sure to strictly define only one node to use as the source, otherwise it will be confusing.
Note: You have to be careful not to include nodes that would link your path to another one. For example,
project are linked to more than one experimental design.
The first two lines are used to select one single node from which to expand. The third line expands the path using these parameters:
n, the starting node or nodes (preferably one for your first queries).
"", the relationship filter. We are not filtering by any relations in this query.
"-project|-protocol", the label filter. We are excluding (hence the minus sign) any nodes with the labels
projector (that is represented by the
0is the minimum depth. Normally 0. Otherwise the starting nodes get excluded.
-1is used to determine the maximum depth for the path expansion. -1 means no limit. If you would set a 1 here, the result would be the
CBTM-376Cdonor and its first level neighbours.
MATCH (n:donor_organism) WHERE n.`biomaterial_core.biomaterial_id` = "CBTM-376C" CALL apoc.path.expand(n, "", "-project|-protocol", 0, -1) YIELD path RETURN path
List of Ingest submission IDs
- KidneySingleCellAtlas: d5410c6e-612d-421a-a66f-2de5e04dd050
- HumanColonicMesenchymeIBD: c51efa69-a348-46d4-9997-c7a11915d53b
- scRNAseqSystemicComparison: d1610c4a-76c6-4b69-af63-c74af869fa75
- TissueStability: fd52efcc-6924-4c8a-b68c-a299aea1d80f
- demo HPSI human cerebral organoids: 6f33a6c6-170d-460b-bbba-9668e9d77aaf
- Mouse Melanoma: 4756b97c-3666-4ff0-9c48-611dc06b7740
- Multiplexed scRNA-seq with barcoded antibodies: 6931d821-9cd2-4ae1-acf7-62cd3b90082e
- Reprogrammed_Dendritic_Cells: 8c45a848-ab32-4928-8dd6-567b75eaf7e1
- HPSI human cerebral organoids: fce97270-fce0-4744-8a4e-a93d95521852
- Tabula Muris: 14df1f92-155c-4da2-97fc-85601dee64da
- snRNA-seq_for_human_retina: 7d515499-b5b6-4083-95cd-6a393344865e
- Healthy and type 2 diabetes pancreas: c81f7d54-a27f-4212-a6df-88dde947f7cc
- Single cell transcriptome analysis of human pancreas: 5cd8b827-e4fa-4ffa-86f5-f0f8302baee9
- SingleCellLiverLandscape: 1e2601a5-8938-446c-bbb3-1f37a84b11da
- WongAdultRetina: 668791ed-deec-4470-b23a-9b80fd133e1c
- Single cell RNAseq characterization of cell types produced over : 0948a727-228f-4cfc-857e-6243c6aed08d
- Kidney biopsy scRNA-seq: 588281cb-93e2-4888-b75e-5cee5e39fc7f
- BM_PC: 67abd7a1-644b-48d0-b8b8-251d8195a742
- Human Hematopoietic Profiling: 7dbcf5ae-f8d7-487c-a3d4-794d8639a1e2
- Mouse Endoderm Project: 0fb44736-c50f-49ab-bcdc-0985e596b955
- Drop-seq, DroNc-seq, Fluidigm C1 Comparison: bf3116e5-1af1-46c2-8bbd-44dac49d1e7f
- 1M Immune Cells: 85e72912-9f91-4489-8169-3b43cc65a16a
- 1M Neurons: 064d36ca-ea4d-428f-b30f-0cf5e5350a9d
- MouseGastrulationAtlas: 19c5c89a-7bfb-4b4d-9ff5-3ef009304ba5
- Fetal/Maternal Interface: 46fe8bd7-329b-4f09-b227-5ee48c109c16
- Tissue stability: 02e89f20-84c8-4daa-aaeb-80f4a85733ff
- CD4+ cytotoxic T lymphocytes: 3bb965cb-2a6f-4802-a4f2-52d0423e6ecc
- Diabetic Nephropathy snRNA-seq: a69c731b-9db7-45e3-938f-daf2e4636a6e
- HDCA-Sweden-10x: 071fc37f-1a01-4cf7-a6f5-2662c42c12b6
- HumanTissueTcellActivation: 8b5feb5e-9039-4c54-9e79-053e490c141a
- HumanMousePancreas: 7b4cd093-bfa5-477e-9c95-69bafc1cb6bf
Releasing a new version to PyPI
You should have a maintainer role access for this project in PyPI.
- Bump the version and create a tag
- Upload to PyPI
This package was created with Cookiecutter.
Release history Release notifications | RSS feed
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size ingest_graph_validator-0.6.3-py2.py3-none-any.whl (27.6 kB)||File type Wheel||Python version py2.py3||Upload date||Hashes View|
|Filename, size ingest-graph-validator-0.6.3.tar.gz (5.1 MB)||File type Source||Python version None||Upload date||Hashes View|
Hashes for ingest_graph_validator-0.6.3-py2.py3-none-any.whl
Hashes for ingest-graph-validator-0.6.3.tar.gz