5 Key Takeaways

• 1

  The 2-step method predicts compound elution order first, then converts that order into retention times for a specific laboratory method.

• 2

  Its graph neural network uses molecular structure, column information, and mobile-phase pH to predict retention order.

• 3

  A polynomial calibration uses confidently identified compounds as anchors; the researchers found that as few as 15 anchors could be sufficient.

• 4

  Across 171 reversed-phase chromatography datasets, the condition-aware model had a retention-order error rate of 4.89%, versus 5.61% without condition information.

• 5

  The method supported compound annotations in human fecal samples but is limited to reversed-phase chromatography and depends on accurate metadata and anchor identifications.