Using predicate and provenance information from a knowledge graph for drug efficacy screening.

BACKGROUND: Biomedical knowledge graphs have become important tools to computationally analyse the comprehensive body of biomedical knowledge. They represent knowledge as subject-predicate-object triples, in which the predicate indicates the relationship between subject and object. A triple can also contain provenance information, which consists of references to the sources of the triple (e.g. scientific publications or database entries). Knowledge graphs have been used to classify drug-disease pairs for drug efficacy screening, but existing computational methods have often ignored predicate and provenance information. Using this information, we aimed to develop a supervised machine learning classifier and determine the added value of predicate and provenance information for drug efficacy screening. To ensure the biological plausibility of our method we performed our research on the protein level, where drugs are represented by their drug target proteins, and diseases by their disease proteins. RESULTS: Using random forests with repeated 10-fold cross-validation, our method achieved an area under the ROC curve (AUC) of 78.1% and 74.3% for two reference sets. We benchmarked against a state-of-the-art knowledge-graph technique that does not use predicate and provenance information, obtaining AUCs of 65.6% and 64.6%, respectively. Classifiers that only used predicate information performed superior to classifiers that only used provenance information, but using both performed best. CONCLUSION: We conclude that both predicate and provenance information provide added value for drug efficacy screening.

Microarray analysis on Runx2-deficient mouse embryos reveals novel Runx2 functions and target genes during intramembranous and endochondral bone formation.

A major challenge in developmental biology is to correlate genome-wide gene expression modulations with developmental processes in vivo. In this study, we analyzed the role of Runx2 during intramembranous and endochondral bone development, by comparing gene expression profiles in 14.5 dpc wild-type and Runx2 (-/-) mice. A total of 1277, 606 and 492 transcripts were found to be significantly modulated by Runx2 in calvaria, forelimbs and hindlimbs, respectively. Bioinformatics analysis indicated that Runx2 not only controls the processes of osteoblast differentiation and chondrocyte maturation, but may also play a role in axon formation and hematopoietic cell commitment during bone development. A total of 41 genes are affected by the Runx2 deletion in both intramembranous and endochondral bone, indicating common pathways between these two developmental modes of bone formation. In addition, we identified genes that are specifically involved in endochondral ossification. In conclusion, our data show that a comparative genome-wide expression analysis of wild-type and mutant mouse models allows the examination of mutant phenotypes in complex tissues.

Gene array analysis of bone morphogenetic protein type I receptor-induced osteoblast differentiation.

UNLABELLED: The genomic response to BMP was investigated by ectopic expression of activated BMP type I receptors in C2C12 myoblast using cDNA microarrays. Novel BMP receptor target genes with possible roles in inhibition of myoblast differentiation and stimulation of osteoblast differentiation were identified. INTRODUCTION: Bone morphogenetic proteins (BMPs) have an important role in controlling mesenchymal cell fate and mediate these effects by regulating gene expression. BMPs signal through three distinct specific BMP type I receptors (also termed activin receptor-like kinases) and their downstream nuclear effectors, termed Smads. The critical target genes by which activated BMP receptors mediate change cell fate are poorly characterized. MATERIALS AND METHODS: We performed transcriptional profiling of C2C12 myoblasts differentiation into osteoblast-like cells by ectopic expression of three distinct constitutively active (ca)BMP type I receptors using adenoviral gene transfer. Cells were harvested 48 h after infection, which allowed detection of both early and late response genes. Expression analysis was performed using the mouse GEM1 microarray, which is comprised of approximately 8700 unique sequences. Hybridizations were performed in duplicate with a reverse fluor labeling. Genes were considered to be significantly regulated if the p value for differential expression was less than 0.01 and inverted expression ratios per duplicate successful reciprocal hybridizations differed by less than 25%. RESULTS AND CONCLUSIONS: Each of the three caBMP type I receptors stimulated equal levels of R-Smad phosphorylation and alkaline phosphatase activity, an early marker for osteoblast differentiation. Interestingly, all three type I receptors induced identical transcriptional profiles; 97 genes were significantly upregulated and 103 genes were downregulated. Many extracellular matrix genes were upregulated, muscle-related genes downregulated, and transcription factors/signaling components modulated. In addition to 41 expressed sequence tags without known function and a number of known BMP target genes, including PPAR-gamma and fibromodulin, a large number of novel BMP target genes with an annotated function were identified, including transcription factors HesR1, ITF-2, and ICSBP, apoptosis mediators DRP-1 death kinase and ZIP kinase, IkappaB alpha, Edg-2, ZO-1, and E3 ligase Dactylin. These target genes, some of them unexpected, offer new insights into how BMPs elicit biological effects, in particular into the mechanism of inhibition of myoblast differentiation and stimulation of osteoblast differentiation.