Editors' ChoiceComputational Biology

Tell me your neighbors, and I will tell you what you are

+ See all authors and affiliations

Science Translational Medicine  08 Feb 2017:
Vol. 9, Issue 376, eaam6058
DOI: 10.1126/scitranslmed.aam6058

Abstract

Network neighbors improve yeast to human gene mapping for the study of parkinsonism.

Yeast or other organisms are often used as a model to study human diseases. To translate the gene functions that are discovered in yeast to human diseases, genes in humans with the most similar sequences to the yeast gene are selected with the hope that this sequence homology captures the human gene(s) with the same function. However, differences at key sites in the genetic sequence may alter a protein’s interaction partners and its function.

Khurana et al. developed a computational approach, TransposeNet, which incorporates gene sequence information alongside structural alignments and information about its neighborhood in genetic and physical interaction networks. Others have incorporated network neighborhoods to map orthologous genes more specifically; however, the authors also used networks to capture orthologs more completely. Using sequence homology alone, 4023 yeast genes could be mapped to 7248 human genes. After the TransposeNet mapping, 4923 proteins were connected to 15,200 human proteins.

TransposeNet mapping allowed the authors to link results from a genome-wide yeast screen against α-synuclein (α-syn) toxicity to human proteins. The α-syn protein forms aggregates associated with Parkinson’s disease and is expected to play a role in the broader class of disorders with similar phenotypes, termed parkinsonism. The authors used the resulting network to identify opportunities to reduce the toxic effects of α-syn associated with parkinsonism. For example, they identified that the overexpression of certain genes suppressed α-syn toxicity in yeast. In human induced pluripotent stem cell–derived neuronal cells expressing mutant α-syn, transcriptionally up-regulating the TransposeNet-mapped orthologs suppressed cellular defects in bulk protein translation.

This work layers structural and network-based information on top of sequence homology to identify the most appropriate human genes to prioritize candidates for functional study. After mapping, the screen results captured a subnetwork that also contains numerous genes linked, in various ways, with parkinsonism. Work to characterize the approach—for example, testing how complete networks must be and benchmarking the method against alternative approaches—still remains, but these results provide a promising foray into studying how genes’ network neighborhoods can be used to more accurately translate findings from model organism screens into disease insights.

Highlighted Article

View Abstract

Related Content

Navigate This Article