Identify unsuitable MRM peaks
Task spaceIdentify peptide peaks that are unsuitable for reliable protein quantification in a liquid chromatography–multiple reaction monitoring (LC–MRM) batch.
Targeted protein measurements depend on distinguishing genuine peptide signals from interference and unreliable peaks. Reviewing each peak in isolation can miss problems that become apparent only when comparing related signals, standards, and samples across a batch.
Agents examine recorded chromatograms for all measured targets in acquisition order and assign each peak a probability of being unsuitable for quantification. Their judgments are compared with the analyst’s recorded labels. The task tests interpretation of analytical measurements and the ability to use batch-wide context to assess individual peaks.
Scoring
Achieve an F1 score of at least 80% on peak rejection decisions against the analyst's labels, balancing precision and recall within each batch.
RESULTS
| Model | Pass@1 | Cost ($)Model/API inference cost only, averaged per task attempt. Lab and labor costs are not reported. Repeats are averaged using the same task and family weights as scores. | |
|---|---|---|---|
| 4.4% | $3.00 | ||
| 0.0% | $12.67 | ||
| 0.0% | $2.49 | ||
| 0.0% | $8.19 | ||
| 0.0% | $15.71 | ||
| 0.0% | $2.60 | ||
0%50%100% | |||
Whiskers show ±1 standard error from repeats.
Model trajectories
One example per model.
Run grading
| Measurement | Recorded result | Outcome |
|---|---|---|
| F1 | 88.9 % | ✓ Pass |
Pass condition: Achieve an F1 score of at least 80% on peak rejection decisions against the analyst's labels, balancing precision and recall within each batch.
Model activity
Loading transcript…