Download Files - KGX Storage

Download knowledge graph files via HTTPS or S3. Both methods are publicly accessible without authentication.

URL behavior

File URLs use the path to the file (e.g. https://kgx-storage.rtx.ai/releases/alliance/latest/graph-metadata.json). Requesting that URL returns the file: JSON is returned as the response body, other formats trigger a download.

For JSON files, appending ?view to the same URL (e.g. .../graph-metadata.json?view) opens the HTML viewer in the browser instead of raw JSON. Only ?view is significant; other query parameters are ignored. Redirects use the canonical path with no query string.

Directory URLs use a trailing slash (e.g. .../latest/). If you request a directory path without a trailing slash, you are redirected to the same path with a trailing slash. Paths that are neither a file nor a directory return 404.

HTTPS Download

Single File

curl -fL -O "https://kgx-storage.rtx.ai/releases/go_cam/latest/go_cam.tar.zst"

Replace go_cam with your source name

Specific Version

curl -fL -O "https://kgx-storage.rtx.ai/data/ctd/November_2025/1.0/normalization_2025sep1/merged_nodes.jsonl"

With wget

wget "https://kgx-storage.rtx.ai/releases/alliance/latest/alliance.tar.zst"

Understanding curl Flags

The recommended curl command uses -fL flags to ensure reliable file downloads:

-L (Follow redirects): The server may use HTTP redirects to route requests to the optimal file location. Without this flag, curl saves the redirect response (typically an HTML page) instead of following the redirect to download the actual file.
-f (Fail on errors): Returns a non-zero exit code when the server responds with an HTTP error (404, 500, etc.). Without this flag, curl silently saves error pages as if they were valid files, which can corrupt your dataset.
-O (Save with remote filename): Saves the file using the same name as on the server.

Using curl -fL ensures your download scripts fail safely and prevent corrupted data from entering your analysis pipeline.

S3 Download

Requires AWS CLI installed locally

Install AWS CLI

brew install awscli

macOS

sudo apt install awscli

Ubuntu/Debian

Single File

aws s3 cp s3://translator-ingests/releases/go_cam/latest/go_cam.tar.zst . --no-sign-request

No AWS credentials required with --no-sign-request

Entire Directory (Recursively)

aws s3 sync s3://translator-ingests/releases/alliance/latest/ ./alliance/ --no-sign-request

Downloads all files in directory

List Available Files

aws s3 ls s3://translator-ingests/releases/ --no-sign-request

Common Paths

Latest Release

releases/{source}/latest/{source}.tar.zst

Merged Files

data/{source}/{version}/{transform}/normalization_{norm}/merged_nodes.jsonl data/{source}/{version}/{transform}/normalization_{norm}/merged_edges.jsonl

Metadata

data/{source}/latest-build.json releases/{source}/latest/graph-metadata.json

Extract Archives

Install zstd

brew install zstd

macOS

sudo apt install zstd