Modern mass spectrometry can detect thousands of molecules in nature, but identifying them all remains a challenge. Researchers must predict chemical structures from tandem mass spectra, even for molecules never seen before.

Current methods mostly compare spectra against reference libraries. They work for known compounds but struggle with the many unknown molecules found in real biological samples. Many still require slow, expensive lab work to identify.

In this competition, you’ll build machine learning models that predict 2D chemical structures (a SMILES string) from LC-MS/MS spectra. Your goal is to generate accurate SMILES representations for both known and novel molecules.

Your solution could help researchers discover new medicines and identify disease biomarkers.