Generating automated kidney transplant biopsy reports combining molecular measurements with ensembles of machine learning classifiers

We previously reported a system for assessing rejection in kidney transplant biopsies using microarray‐based gene expression data, the Molecular Microscope^® Diagnostic System (MMDx). The present study was designed to optimize the accuracy and stability of MMDx diagnoses by replacing single machine learning classifiers with ensembles of diverse classifier methods. We also examined the use of automated report sign‐outs and the agreement between multiple human interpreters of the molecular results. Ensembles generated diagnoses that were both more accurate than the best individual classifiers, and nearly as stable as the best, consistent with expectations from the machine learning literature. Human experts had ≈93% agreement (balanced accuracy) signing out the reports, and random forest‐based automated sign‐outs showed similar levels of agreement with the human experts (92% and 94% for predicting the expert MMDx sign‐outs for T cell–mediated (TCMR) and antibody‐mediated rejection (ABMR), respectively). In most cases disagreements, whether between experts or between experts and automated sign‐outs, were in biopsies near diagnostic thresholds. Considerable disagreement with histology persisted. The balanced accuracies of MMDx sign‐outs for histology diagnoses of TCMR and ABMR were 73% and 78%, respectively. Disagreement with histology is largely due to the known noise in histology assessments (ClinicalTrials.gov NCT01299168).     

State of the art and new developments in molecular diagnostics for hemoglobinopathies in multiethnic societies

For detecting carriers of thalassemia traits, the basic part of diagnostics consists of measurement of the hematological indices followed by mostly automatic separation and measurement of the Hb fractions, while direct Hb separation either on high pressure liquid chromatography or capillary electrophoresis is sufficient to putatively identify carriers of the common Hb variants like HbS, C, E, D, and O‐Arab. A putative positive result is reported together with an advice for parents, partner, or family analysis. For couples, presumed at‐risk confirmation at the DNA level is essential. In general, this part of diagnostics is done in specialized centers provided with sufficient experience and the technical tools needed to combine hematological and biochemical interpretation with identification of the mutations at the molecular level. State‐of‐the‐art tools are usually available in centers that also provide prenatal diagnosis and should consist of gap‐PCR for the common deletions, direct DNA sequencing for all kind of point‐mutations and the capacity to uncover novel or rare mutations or disease mechanisms. New developments are MLPA for large and eventually unknown deletion defects and microarray technology for fine mapping and primer design for breakpoint analysis. Gap‐PCR primers designed in the region flanking the deletion breakpoints can subsequently be used to facilitate carrier detection of uncommon deletions in family members or isolated populations in laboratories where no microarray technology or MLPA is available.     

Oncogenomics: prospects for the future

Genomics has generated a wealth of data that is now being used to identify additional molecular alterations associated with cancer development. Mapping these alterations in the cancer genome is a critical first step in dissecting oncological pathways. There are two ways in which cancer research has changed in recent years. The first is the progressive elucidation of the genomic basis of cancer. This has been accomplished by the generation of detailed information using procedures such as global expression profiling. The second is a renewed emphasis on the role of epigenetic modifications in the etiology of cancer. Changes in DNA methylation and chromatin modification patterns are some of the epigenetic factors that cause gene deregulation in cancer. In this article, current and evolving genomic applications and the hypotheses underlying the modality for cancer therapy will be reviewed.