Overview
ROASMI learns molecular representations and pairwise elution preferences using a neural ranking model. Its supplied scripts predict retention scores, optionally retrain models, and integrate retention evidence with MS/MS candidate scores.
What can this tool do?
Common uses
- Add chromatographic evidence to molecular candidate ranking.
- Retrain a retention-order model with a suitable reference set.
Scientific fit
Primary function
Metabolite annotation and identification
Data compatibility
Inputs
Compound list or annotations, Feature or peak table
Outputs
Compound list or annotations
Input formats
CSV or TSV
Output formats
Predicted retention scores and candidate annotation rankings
Scope and considerations
Designed for
- Predict relative LC retention behavior and combine it with MS/MS scores to rank candidate structures.
Not designed for
- Replacing experimental confirmation of compound identity.
- Raw-data feature detection or chromatographic alignment.
Important requirements
- Python/Conda and the documented dependencies, model files and candidate structures.
- Prepare input CSV files using the repository prediction and identification examples. Identification requires the expected retention and candidate-score information.
- Select reference data and models compatible with the chromatographic mode and buffer pH.
Considerations
- The supplied initial models target reversed-phase LC around pH 2.7; they are not universal retention models.
- The repository supplies retraining examples, including HILIC. That does not establish general applicability of the initial RPLC models to HILIC.
- CPU execution is documented; GPU acceleration is recommended for faster retraining, not required for all use.
- Retention scores describe relative order, not necessarily calibrated absolute retention times.
Access and documentation
MIT license verified in the official repository. Software was not installed or execution-tested; dependencies and external services have their own requirements.
License: MIT
Evidence and review
Enveda intake reviewed against the primary publication, official repository documentation and license on 2026-10-05. Editorial review only; no execution or independent benchmark.
Core publications
Resource papers, direct updates and tool-specific methods. Selection policy.
- ROASMI: accelerating small molecule identification by repurposing retention data. Journal of Cheminformatics (2025). doi:10.1186/s13321-025-00968-8
Related tools
Entry history
Changes to this catalogue entry, not the tool's software release history. Acceptance and editorial review are separate.
- 2026-10-06 · First committed entry
Add annotation software and link existing Strategy implementations
View change · Entry at this revision