To reduce the rate of false positive target site predictions, we applied the three criteria that were proposed by Hsu and colleagues [54]: target sites must be predicted by at least two tools, they must be located in accessible regions, and the target genes must contain multiple target sites.
For adult exposure predictions, we applied self-reported body weights obtained at the time of survey throughout adulthood.
To demonstrate how our system can yield practical predictions, we applied it repeatedly to the glycolysis and pentose phosphate pathways of S.cerevisiae, to see which new reactions and possibly new compounds would be predicted.
To test whether ET rank smoothness correlates with the quality of functional site predictions, we applied the Laplacian operator to the ET rank distributions on 74 diverse proteins bound to various substrates, cofactors, DNA or proteins (see 'Methods' section).
To experimentally test those predictions, we applied different manipulations designed to force neurons in one or the other direction across their tipping point, therein acutely reproducing neuropathic excitability in primary afferent neurons from naive animals or acutely reversing neuropathic excitability in neurons from nerve-injured animals.
To assess the quality of the predictions, we applied the three new versions of HoxPred on a large set of 800 homeodomain sequences from 9 non-vertebrate species spanning various bilaterian phylogenetic groups, and which include well-characterized Hox and ParaHox genes (Additional file 2, Tables S1, S2, S3).
To evaluate the proposed approach in improving LCC prediction, we apply the proposed uncertainty propagation approach on the LCC model described by Qiang and Lam in [40] to the Saint-Denis city, Reunion Island.
In order to estimate the impact of each class on 3-state solvent accessibility prediction, we apply each of them to train the model and perform the prediction.
Given the very high dimensionality of the VBM output (thousands of voxels, or features, for each subject, each one corresponding to one dimension) and the expectation that only a few of these features would be meaningful for prediction, we applied a further feature selection step [36].
To select the genes for prediction, we applied an approach based upon genetic algorithms (GA).
