PreprintCC BY 4.0arXiv
Nearest-neighbour baselines for fingerprint prediction from MS/MS spectra under different assumptions
Abstract
It has recently been shown that nearest-neighbour retrieval provides a strong baseline for molecular fingerprint prediction from MS/MS spectra, with several variants matching or outperforming current deep learning models (Khoo and Barzilay, 2026; Liu et al., 2026; Gupta et al., 2026). Importantly, "nearest neighbour" encompasses a family of retrieval methods that differ in the information assumed to be available at inference. In this report, we systematically compare several nearest-neighbour variants and show how these differing assumptions affect performance. Our goal is to establish stricter baselines that enable more rigorous benchmarking and better measure progress in this area.
Related
- Prediction of the performance of industrial gas turbine based on different environmental conditions
- A law of thin processes with neighbour-count thinning
- Spatial Functional k-Nearest-Neighbour Regression under Polynomial Dependence
- Gaussian and non-Gaussian limits for the total edge-length of random directional nearest-neighbour graphs
- Scalene Yang-Baxter triples realize all nearest-neighbour Hamiltonians with global additive symmetries
BibTeX
@misc{khoo2026nearestneighbour,
title = {{Nearest-neighbour baselines for fingerprint prediction from MS/MS spectra under different assumptions}},
author = {Ling Min Serena Khoo},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.02249},
eprint = {2610.02249},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.02249},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Nearest-neighbour baselines for fingerprint prediction from MS/MS spectra under different assumptions
AU - Ling Min Serena Khoo
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.02249
UR - https://arxiv.org/abs/2610.02249
AB - It has recently been shown that nearest-neighbour retrieval provides a strong baseline for molecular fingerprint prediction from MS/MS spectra, with several variants matching or outperforming current deep learning models (Khoo and Barzilay, 2026; Liu et al., 2026; Gupta et al., 2026). Importantly, "nearest neighbour" encompasses a family of retrieval methods that differ in the information assumed to be available at inference. In this report, we systematically compare several nearest-neighbour variants and show how these differing assumptions affect performance. Our goal is to establish stricter baselines that enable more rigorous benchmarking and better measure progress in this area.
ER -