The SVMs were trained using the target person's feature vectors as positive examples (25 samples for training) and the non-target persons' feature vectors as negative examples (275 samples for training).
Cells were also infected with corresponding vectors as negative control.
Agglomerative clustering was implemented using the aggExCluster method in the apcluster R package with the mutual pairwise similarities of data vectors as negative distances [ 38].
In initial experiments we found that transduction of the cells with control vectors encoding either AP or Hyal1 had no effect on cell morphology (data not shown), and we used these vectors as negative controls for the effects of vector transduction in subsequent experiments.
Culture supernatant from a yeast clone transformed with empty pPIC9 vector served as negative control.
Concentrated culture supernatant from a yeast clone transformed with empty pPIC9 vector served as negative control.
The empty vector served as negative control in all experiments.
Cells transfected with an empty vector served as negative controls in these studies.
Reaction mixtures containing microsomes prepared from WAT11 transformed with the empty vector served as negative controls.
We used the empty vector pGR106 as negative control, which upon transient expression showed no micro- and/or macroscopic phenotype.
