Design of Biomedical Robots for Phenotype Prediction Problems

Abstract

Genomics has been used with varying degrees of success in the context of drug discovery and in defining mechanisms of action for diseases like cancer and neurodegenerative and rare diseases in the quest for orphan drugs. To improve its utility, accuracy, and costeffectiveness optimization of analytical methods, especially those that translate to clinically relevant outcomes, is critical. Here we define a novel tool for genomic analysis termed a biomedical robot in order to improve phenotype prediction, identifying disease pathogenesis and significantly defining therapeutic targets. Biomedical robot analytics differ from historical methods in that they are based on melding feature selection methods and ensemble learning techniques. The biomedical robot mathematically exploits the structure of the uncertainty space of any classification problem conceived as an ill-posed optimization problem. Given a classifier, there exist different equivalent small-scale genetic signatures that provide similar predictive accuracies. We perform the sensitivity analysis to noise of the biomedical robot concept using synthetic microarrays perturbed by different kinds of noises in expression and class assignment. Finally, we show the application of this concept to the analysis of different diseases, inferring the pathways and the correlation networks. The final aim of a biomedical robot is to improve knowledge discovery and provide decision systems to optimize diagnosis, treatment, and prognosis. This analysis shows that the biomedical robots are robust against different kinds of noises and particularly to a wrong class assignment of the samples. Assessing the uncertainty that is inherent to any phenotype prediction problem is the right way to address this kind of problem.

Genomic data integration in chronic lymphocytic leukemia

“…In this case, we used a two‐dimensional plot to show that the classification problem approaches a linearly separable behavior in the PCA space (it is possible to separate both populations by a plane) when the dataset is reduced to the corresponding small‐scale signatures, in contrast that observed when all the genetic probes are taken into account. This can be considered as graphical proof of the discriminatory power of these small‐scale signatures . Figures and show the PCA plots for IgVH , NOTCH1 and SF3B1 mutational status.…”

“…The cumulative distribution function of the small‐scale predictive accuracies found in different hold‐outs is finally presented and serves to account for the variability in its predictive accuracy with partial information. A statistical analysis is performed providing the minimum, maximum and median bounds that could be expected in an independent dataset. Sampling the uncertainty space in the corresponding phenotype prediction problems to perform pathways analysis . In this procedure, we look for the most predictive genes found in each random hold‐out.…”

The Effect of NOP16 Mutation in Chronic Lymphocytic Leukemia

“…The methodology tries to determine the shortest lists of most discriminatory genes that predict the NOP16 mutation and is described by Fernández-Martínez et al [20] and De Andrés-Galiana et al [22][23][24]. This classification problem is naturally unbalanced due to the low number of patients that show the NOP16 mutation, and the classifier has to take this feature into account.…”

“…We have used the Fisher's Ratio and Fold Change to rank the genes according to their discriminatory power [20,[22][23][24]. These gene-ranking methods and particularly Fisher's ratio turned to be very robust against different kind of noise [24].…”

“…It is important to understand that the phenotype prediction problems admit different genetic signatures explaining equally well the phenotype, due to the reduced number of samples that are available compared to the number of genetic probes that are monitored [23]. Therefore, the pathway analysis is sensible to this high degree of under determinacy.…”