Skip to content

Outputs

Autonima writes run artifacts under the resolved runtime output folder.

If you run:

autonima run projects/cue_reactivity/default.yaml

the default output folder is:

projects/cue_reactivity/default/

Typical Layout

<output-folder>/
├── outputs/
│   ├── search_results.json
│   ├── abstract_screening_results.json
│   ├── fulltext_retrieval_results.json
│   ├── fulltext_screening_results.json
│   ├── execution_manifest.json
│   ├── execution_progress.json
│   ├── config.executed.yaml
│   ├── final_results.json
│   ├── missing_fulltexts.txt
│   ├── missing_fulltexts.csv
│   ├── criteria_mapping.json
│   ├── coordinate_parsing_results.json
│   ├── nimads_studyset.json
│   └── nimads_annotation.json
├── retrieval/
│   └── pubget_data/
└── meta_analysis_results/

The exact contents depend on which stages are enabled.

Need Help Interpreting These Files?

Use the Interpreting Outputs guide for a task-oriented walkthrough of what each artifact means and which file to inspect first.

Key Files

outputs/search_results.json

  • search-stage results
  • study metadata from PubMed or provided PMIDs

outputs/abstract_screening_results.json

  • decisions and reasoning from abstract screening

outputs/fulltext_retrieval_results.json

  • retrieval summary and availability information

outputs/fulltext_screening_results.json

  • decisions and reasoning from full-text screening

outputs/final_results.json

  • final included/excluded status and execution outputs

Execution and cache records

  • execution_manifest.json records the exact semantic hashes and cache plan for the run.
  • execution_progress.json records per-stage status and whether work was fresh, reused, or mixed.
  • config.executed.yaml is the normalized runtime config snapshot used for that output folder.

outputs/missing_fulltexts.txt

  • PMIDs where full-text screening returned fulltext_incomplete

outputs/missing_fulltexts.csv

  • PMIDs where full-text screening returned fulltext_incomplete
  • includes full_text_path to help locate/fix the problematic source text

outputs/criteria_mapping.json

  • generated mapping of screening criteria IDs used in downstream results

outputs/coordinate_parsing_results.json

  • cached coordinate parsing results when parsing is enabled

outputs/nimads_studyset.json

outputs/nimads_annotation.json

  • NiMADS artifacts used by autonima meta
  • pass the containing folder to autonima meta, usually <output-folder>/outputs

retrieval/pubget_data/

Contains retrieved article data from PubGet, such as:

  • metadata.csv
  • text.csv
  • coordinates.csv
  • extracted table data

meta_analysis_results/

Created by autonima meta. Contains one directory per annotation column and the generated NiMARE artifacts and reports.

Practical Notes

  • The CLI runtime output path may differ from the output.directory stored in YAML.
  • If you rerun with the same output folder, Autonima reuses only signed entries whose stage settings and per-study inputs still match.
  • Final exports are regenerated even when every upstream decision is reused.
  • When documenting or sharing a run, include both the config file and the resolved runtime output folder.