dockerPractice

Practice Using Docker on git

Main Link

https://www.arothuis.nl/posts/one-off-docker-images/

Goal

Given an a custom file framework of Nodes and strings to map for those nodes use a few doc APIs to map the strings to Sense.ID and create several audit files and intermediary files to parse. This will be a stand alone application using docker.

Inputs

A dict of dicts called api_empty_dict.json. E.G. {'Disease': {'type 2 diabetes mellitus', 'sarcoma', 'gout'}}

Process

Get the api_empty_dict.json and query the two APIs

• Analytics: https://caladan.doctorevidence.com/portal/suggestions?search={stroke}
• Search: https://search.doctorevidence.com/api/annotator/batch-annotate

Next the apis are compared to see if they are in agreement. Ideally each string returns only 1 Sense.ID and both APIs return the same one. This is not always the case and a series of assumptions are made and documented to get the 2 APIs more inline. Finally a report, appendix, and audit file is created.

Output Schema

The output folder cintains 2 pdfs one for the report and one for the appendix tables of audit files. The rawData contains the filled in api_empty_dict.json from the api result. The parsedData folder contains a formated Data.Frame that is used in analysis. The images folder contains the images inserted into the pdf.

Output
├──report.pdf
├──Appendix.pdf
├──apiAuditFile.pkl
├── rawData
│   ├── analyticsAPI_dict.pkl
│   ├── searchAPI_dict.pkl
├── parsedData
│   ├── analyticsAPI_DF.pkl
│   ├── searchAPI_DF.pkl
├── images
│   ├── filename*.png

Filetypes

*.pkl: This is just an internal python data dump file that can be read with the pickle library.

Call

Docker use local folders for inputs and outputs passed as volumes.
The input folder should contain a file called api_empty_dict.json The output folder should be empty.

Example call: docker run -v $pwd/output:/output -v $pwd/input:/input example1

TBD

• The outputs currently expect 9 nodes to create a 3x3.