'_samples' is a required property: (1 samples) project

rschauner commented 1 month ago

I'm trying to create a PEP schema for a Snakemake workflow I'm writing. I've parsed my config file and sample table to the smallest form I can and still get the same error: '_samples' is a required property: (1 samples) project.

config.yml

pep_version: 2.1.0
sample_table: "test_table.csv"

test_table.csv

sample_name, exp
P081-BAL-NonInf-Multi, A
P081-BAL-InfAd-Multi, A
P081-BAL-InfAdPep-Multi, B

schema.yml

imports:
  - http://schema.databio.org/pep/2.1.0.yaml
properties:
  config:
    properties:
    required:
      - pep_version
      - sample_table
  samples:
    type: array
    items:
      type: object
      properties:
        exp:
          type: string
          description: "Experimental Condition"
      required:
        - sample_name
        - exp

When I run eido validate pep/config.yml -s http://schema.databio.org/pep/2.1.0.yaml I get the error. However, running eido with either the -c or -n flags validates the PEP.

Removing the last two lines (see below) of the PEP schema removes this error. Changing the "sample_name" column to "sample_name2" results in an error: sample_table is missing 'sample_name' column; you must specify sample_tables in sample_name or derive them. Using a sample CSV with a schema that requires a metadata column still validates with that column missing (using eido validate pep/config.yml -s schema.yml