SegmentationLoader

Load WSI segmentations to mongodb instance in a Docker network environment (i.e. QuIP).

Setup

Folder Structure

See EXAMPLE.

Directory structure of ./data:

├── data/
│   ├── segmentation_results/
|       ├── manifest.csv
│       ├── TCGA-A2-A3XZ-01Z-00-DX1.svs/ (folder)
|           ├── ...lots of segmented files
|

The manifest file will on the same level as the directory that contains

Manifest file

Segmentation data is one folder per slide. So in the manifest.csv file, you may have, for example:

segmentdir,studyid,clinicaltrialsubjectid,imageid
TCGA-A2-A3XZ-01Z-00-DX1.svs,TCGA-A2-A3XZ-01Z,TCGA-A2-A3XZ-01Z-00-DX1,TCGA-A2-A3XZ-01Z

Example Usage

Loading segmentations:

Running the loader as a background process is highly recommended.

# Command with arguments:
nohup docker exec quip-segloader loadfiles --src [data_folder] --collectionname [pathdb collection] --user [username] --passwd [password] > /dev/null &

# Example:
nohup docker exec quip-segloader loadfiles --src segmentation_results --collectionname TCGABRCA --user [your username] --passwd [your password] > /dev/null &

The > /dev/null & is for keeping any error logs that may occur, but not generate any other output files (so as to not eat up all the space).

data_folder = top-level directory containing segmentation subfolder and manifest.csv

FYI, the upper level directory (in this case, ./data) is the folder that is mapped to docker.

In this example, the folder name you would pass to the program is segmentation_results.

Build

docker-compose.yml included for convenience

Edit the file and edit volumes - set the source to your main data folder.

Currently, source is set to ./data

If your network name should change (Hint: it shouldn't) but if it does, then change networks as well.