Predicting chemical shifts with graph neural networks

Yang, Zeliang; Chakraborty, Maghesree; White, Andrew D.

doi:10.1039/d1sc01895g

Cited by 42 publications

(71 citation statements)

References 62 publications

(88 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The simulation duration was 1.3 µs with frames saved for this analysis every 500 ps. To compute the chemical shift, g k , we use a graph neural network that can compute chemical shift from atomic positions [73]. We only biased backbone HN atoms, due to their higher accuracy [73].…”

Section: Example System 3: Mbp Fragment Molecular Dynamicsmentioning

confidence: 99%

“…To compute the chemical shift, g k , we use a graph neural network that can compute chemical shift from atomic positions [73]. We only biased backbone HN atoms, due to their higher accuracy [73]. The first 6 HN atoms were biased (NPVVHF), excluding the N-terminus.…”

Section: Example System 3: Mbp Fragment Molecular Dynamicsmentioning

confidence: 99%

“…Prior was reweighted with MaxEnt to have agreeing chemical shifts at indicated sequence positions. Chemical shifts are HN, as predicted by graph neural network[73]. Experiments from Spyranti et al[71].…”

mentioning

confidence: 96%

See 2 more Smart Citations

Simulation-based inference with approximately correct parameters via maximum entropy

Barrett

Ansari

Ghoshal

et al. 2022

Mach. Learn.: Sci. Technol.

Self Cite

View full text Add to dashboard Cite

Inferring the input parameters of simulators from observations is a crucial challenge with applications from epidemiology to molecular dynamics. Here we show a simple approach in the regime of sparse data and approximately correct models, which is common when trying to use an existing model to infer latent variables with observed data. This approach is based on the principle of maximum entropy (MaxEnt) and provably makes the smallest change in the latent joint distribution to ﬁt new data. This method requires no likelihood or model derivatives and its ﬁt is insensitive to prior strength, removing the need to balance observed data ﬁt with prior belief. The method requires the ansatz that data is ﬁt in expectation, which is true in some settings and may be reasonable in all with few data points. The method is based on sample reweighting, so its asymptotic run time is independent of prior distribution dimension. We demonstrate this MaxEnt approach and compare with other likelihood-free inference methods across three systems: a point particle moving in a gravitational ﬁeld, a compartmental model of epidemic spread and ﬁnally molecular dynamics simulation of a protein.

show abstract

Section: Example System 3: Mbp Fragment Molecular Dynamicsmentioning

confidence: 99%

Section: Example System 3: Mbp Fragment Molecular Dynamicsmentioning

confidence: 99%

See 1 more Smart Citation

Simulation-based inference with approximately correct parameters via maximum entropy

Barrett

Ansari

Ghoshal

et al. 2022

Mach. Learn.: Sci. Technol.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Recently a GNN architecture was described that can capture the effect of noncovalent interactions and secondary structure effects on chemical shis of C, N and H nuclei in biomacromolecules and organic molecules. 33 Empirical approaches to NMR chemical shi prediction use interatomic connectivity to dene the local neighborhood around a given atom, while the effects of stereochemistry and molecular conformation are most oen ignored. However, geometric factors play a fundamental role in inuencing chemical shi.…”

Section: Introductionmentioning

confidence: 99%

Real-time prediction of ¹H and ¹³C chemical shifts with DFT accuracy using a 3D graph neural network

et al. 2021

View full text Add to dashboard Cite

show abstract

“…A large focus has been on applying graph neural networks to predicting molecular properties. 1 , 2 , 3 , 4 , 5 , 6 These works assume a central server that has access to all data. However, such a centralized-learning scenario may not represent how institutions share chemical data.…”

Section: Introductionmentioning

confidence: 99%

Federated learning of molecular properties with graph neural networks in a heterogeneous setting

2022

Self Cite

View full text Add to dashboard Cite

Predicting chemical shifts with graph neural networks

Abstract: Inferring molecular structure from Nuclear Magnetic Resonance (NMR) measurements requires an accurate forward model that can predict chemical shifts from 3D structure. Current forward models are limited to specific molecules...

Cited by 42 publications

References 62 publications

Simulation-based inference with approximately correct parameters via maximum entropy

Simulation-based inference with approximately correct parameters via maximum entropy

Real-time prediction of ¹H and ¹³C chemical shifts with DFT accuracy using a 3D graph neural network

Federated learning of molecular properties with graph neural networks in a heterogeneous setting

Contact Info

Product

Resources

About

Predicting chemical shifts with graph neural networks

Abstract: Inferring molecular structure from Nuclear Magnetic Resonance (NMR) measurements requires an accurate forward model that can predict chemical shifts from 3D structure. Current forward models are limited to specific molecules...

Cited by 42 publications

References 62 publications

Simulation-based inference with approximately correct parameters via maximum entropy

Simulation-based inference with approximately correct parameters via maximum entropy

Real-time prediction of 1H and 13C chemical shifts with DFT accuracy using a 3D graph neural network

Federated learning of molecular properties with graph neural networks in a heterogeneous setting

Contact Info

Product

Resources

About

Real-time prediction of ¹H and ¹³C chemical shifts with DFT accuracy using a 3D graph neural network