A grouped split silently removes the chemistry entangled with its grouping variable: the case of MOF topology hold-out
- 1College of Chemical Engineering
- 2Nanjing Tech University
Abstract
Machine learning screens metal–organic frameworks (MOFs) for CO2 capture, and 2026 studies on enumerated databases have converged on topology hold-out as a test of topological transferability, presupposing untested separability of topology and chemistry. They are not: on hMOF (324,426 structures), 74.7% of coarse compositions (61.3% of structures) occur in a single topology, and for seven of hMOF's eleven topologies removing one topology leaves every structure with no chemical counterpart in training; the confound replicates on ARC-MOF (99.3% coarse single-topology), which is not independent of hMOF, so only a leave-one-source-out bound (98.6–99.8%) is claimed. On the isoreticular pcu ↔ nbo pair, removing the matched topology degrades nbo-side MAE by 70.0% while a sizematched control leaves it unchanged, and across the shared 25,000-sample support set the top-1% hit rate falls from 50.4% to 19.3% (five seeds). Yet the topology penalty is not explained by chemistry alone: holding out the composite chemistry key instead retains R2 = 0.674 against topology's 0.379, so the composition counts fix the identity of the withheld chemistry, not its cost. Topology hold-out should therefore not be read as a clean test of topological transferability, and grouped-split reports should first disclose the confound structure and chemical granularity, for which we provide a reporting checklist.
Related
- Topology-Dependent Glass Formation Behavior in Poly[n]catenane Melts with Variable Ring Size and Chain Length
- Multiply entangled receptors for high-affinity anion recognition in water
- A skin-like, all-in-one hydrogel engineered via a supramolecular entangled network reinforced with hydrogen-bond–associated nanodomains
- Constraint-Limited Tube Orientation of Entangled Polymers in Oscillatory Shear Deformation
- Selective Biexciton Generation Under Energy-Time Entangled Quantum Light in Quantum Dots
BibTeX
@misc{wang2026grouped,
title = {{A grouped split silently removes the chemistry entangled with its grouping variable: the case of MOF topology hold-out}},
author = {Tao Wang},
howpublished = {ChemRxiv},
year = {2026},
month = oct,
doi = {10.26434/chemrxiv.15010692/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010692/v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - A grouped split silently removes the chemistry entangled with its grouping variable: the case of MOF topology hold-out
AU - Tao Wang
PY - 2026
DA - 2026/10/20
PB - ChemRxiv
DO - 10.26434/chemrxiv.15010692/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15010692/v1
AB - Machine learning screens metal–organic frameworks (MOFs) for CO2 capture, and 2026 studies on enumerated databases have converged on topology hold-out as a test of topological transferability, presupposing untested separability of topology and chemistry. They are not: on hMOF (324,426 structures), 74.7% of coarse compositions (61.3% of structures) occur in a single topology, and for seven of hMOF's eleven topologies removing one topology leaves every structure with no chemical counterpart in training; the confound replicates on ARC-MOF (99.3% coarse single-topology), which is not independent of hMOF, so only a leave-one-source-out bound (98.6–99.8%) is claimed. On the isoreticular pcu ↔ nbo pair, removing the matched topology degrades nbo-side MAE by 70.0% while a sizematched control leaves it unchanged, and across the shared 25,000-sample support set the top-1% hit rate falls from 50.4% to 19.3% (five seeds). Yet the topology penalty is not explained by chemistry alone: holding out the composite chemistry key instead retains R2 = 0.674 against topology's 0.379, so the composition counts fix the identity of the withheld chemistry, not its cost. Topology hold-out should therefore not be read as a clean test of topological transferability, and grouped-split reports should first disclose the confound structure and chemical granularity, for which we provide a reporting checklist.
ER -