We evaluate the impact of pretraining Graph Transformer architectures on atom-level quantum-mechanical features for the modeling of absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of drug-like compounds. We compare this p
where the individual splits are optimized according to the gradient and the Hessian of some loss function (i.e. cross-entropy), and converts the sum of predictions into a probability by applying the sigmoid function [31]
Lcontrastive=∑i∈batchLi=∑i∈batch−1|P(i)|∑p∈P(i)logexp(zizizpzp/τ)∑a∈A(i)exp(zizizaza/τ) (3) P(i) is a set that includes all samples whose labels are the same as samplei,|∙|is a function that counts the number of elements in a set,ziis a feature vect...
Chelliah, V., Li, L., He, E., Henry, A., Stefan, M., Snoep, J., Hucka, M., Le Nov\(\grave{e}\)re, N., & Laibe, C. (2010). BioModels database: An enhanced, curated and annotated resource for published quantitative kinetic models.BMC Systems Biology, 4, 92. ...