Skip to content

nerd create

nerd create ingests metadata (constructs, buffers, sequencing runs, samples) into a NERD SQLite database. It is typically the first task you run when setting up a new analysis.

nerd create --config PATH/TO/config.yaml --db PATH/TO/nerd.sqlite

If --db is omitted, the CLI creates/updates nerd.sqlite inside run.output_dir.

Configuration Layout

Every config has two top-level sections:

run:
  label: create_all_samples         # used in output paths
  output_dir: examples              # base directory for logs & artifacts
  backend: local                    # optional: slurm / ssh

create:
  buffers: buffers.csv
  constructs: constructs.csv
  sequencing_runs: sequencing_runs.csv
  samples: probing_samples.csv
  derived_samples: derived.csv      # optional
  • run block – standard NERD metadata (label, output directory, execution backend, resources).
  • create block – describes what to import. Values can be inline objects or CSV filenames relative to the config (or resolved via search_roots).

CSV Mode

When a value is a string (e.g., samples: probing_samples.csv), nerd create treats it as a sheet:

  • buffers.csv → inserted into meta_buffers.
  • constructs.csv → inserted into meta_constructs (optionally with nt_info paths).
  • sequencing_runs.csv → inserted into sequencing_runs.
  • probing_samples.csv → inserted into sequencing_samples, plus derived probe_reactions/probe_reaction_groups.

Columns should match the spreadsheet headers (see examples/create_all_samples/to_import/). Extra columns are ignored; missing required columns raise errors.

YAML Mode

You can inline dictionaries instead of CSV files:

create:
  buffers:
    - name: Schwalbe_bistris
      pH: 6.5
      composition: Tris/Bicine
      disp_name: pH6.5_A
  constructs:
    - family: HIV
      name: fourU_new
      version: 2
      sequence: ACTG...
      disp_name: 4U_wt
      nt_info: fourU_wt_nt.csv     # optional relative path

Inline mode is best for small test cases; CSV mode scales better for large studies.

Sample Ingestion Details

When loading samples from CSV:

  • sequencing_run_name is used to look up (or create) sequencing runs.
  • construct and buffer names are resolved against the database (using either disp_name or name). Missing references throw errors before any reactions are inserted.
  • reaction_group labels are turned into probe_reaction_groups, and each sample becomes a row in probe_reactions with the associated metadata (temperature, reaction_time, etc.).

If you omit constructs or buffers in the sheet, you may supply defaults in the YAML; otherwise, nerd create requires every sample to specify a valid reference.

Trace of Inserted Objects

The task writes two log files in the run directory:

  • created_objects.json
  • created_objects.log

Both summarize how many buffers, constructs, sequencing runs, and samples were created or updated. Use them to confirm the import before running downstream tasks.

Common Flags

Option Description
--config Path to YAML file (required).
--db SQLite database path; defaults to <run.output_dir>/nerd.sqlite.
--verbose Enables DEBUG logging.
--log-file Write logs to a custom path.

Because nerd create runs inside the CLI’s task framework, you can also supply --backend, --threads, --mem-gb, etc., in the run section to scale the job.

Examples

# Minimal inline config
nerd create --config examples/create_minimal.yaml --db nerd.sqlite

# Full CSV import (buffers, constructs, sequencing runs, samples)
nerd create --config examples/create_all_samples/create_all.yaml --db nerd.sqlite

# Separate probing sample import
nerd create --config examples/create_all_samples/create_probing_samples.yaml \
            --db nerd.sqlite

Once metadata is ingested, you’re ready to run nmr_* or probe_timecourse tasks on the same database.