Translation of Genotype to Phenotype by a Hierarchy of Cell Subsystems

Yu, Michael; Kramer, Michael H.; Dutkowski, Janusz; Srivas, Rohith; Licon, Katherine; Kreisberg, Jason F.; Ng, Cherie; Krogan, Nevan J.; Sharan, Roded; Ideker, Trey

doi:10.1016/j.cels.2016.02.003

Cited by 73 publications

(100 citation statements)

References 77 publications

(100 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We attribute Ontotype’s relatively poor performance on our human data to the sparsity and incompleteness of functional annotations in human; this finding further highlights the strength of our approach where functional relationships are directly inferred from interactomes as opposed to relying on curated annotations. Additionally, we observed that Mashup achieves better overall performance, albeit by a small margin (AUPR of 0.13 compared to 0.1), than Ontotype on the original yeast data set (Costanzo et al, 2010) used by Yu et al (2016; Figure S9). …”

Section: Resultsmentioning

confidence: 92%

“…Mashup’s performance was highly consistent across a wide range of choices for the dimensionality of our learned representation (Figure S5). Furthermore, Mashup achieved substantially better accuracy than a recent approach that uses known GO annotations of each gene pair as input features for random forest classifiers (Yu et al, 2016; referred to as Ontotype), which was shown to achieve state-of-the-art performance in yeast. We attribute Ontotype’s relatively poor performance on our human data to the sparsity and incompleteness of functional annotations in human; this finding further highlights the strength of our approach where functional relationships are directly inferred from interactomes as opposed to relying on curated annotations.…”

Section: Resultsmentioning

confidence: 99%

“…To evaluate the performance of Ontotype (Yu et al, 2016), we first built a hierarchy of GO terms across all three ontologies (biological process, molecular function, and cellular component) using only “is a” and “part of” relationships. Then, we assigned each gene in STRING to its known GO terms and all of their ancestors in the hierarchy.…”

Section: Star Methodsmentioning

confidence: 99%

“…Given a pair of genes, we constructed a feature vector (termed Ontotype) that has 0, 1, or 2 for each GO term representing the number of genes in the pair that are assigned to the term. Following Yu et al (2016), we used the implementation of random forest classifiers provided by the Python scikit-learn package (Pedregosa et al, 2011) to classify genetic interactions based on the Ontotype features. We explored a wide range of model parameters: {100, 300, 500, 1000} for number of trees, {10, 30, 50, Full} for maximum depth of the trees, and {0.1, 0.3, 0.5} for the fraction of features to consider at each split.…”

Section: Star Methodsmentioning

confidence: 99%

See 3 more Smart Citations

Compact Integration of Multi-Network Topology for Functional Analysis of Genes

Cho¹,

2016

View full text Add to dashboard Cite

Summary The topological landscape of molecular or functional interaction networks provides a rich source of information for inferring functional patterns of genes or proteins. However, a pressing yet unsolved challenge is how to combine multiple heterogeneous networks, each having different connectivity patterns, to achieve more accurate inference. Here we describe the Mashup framework for scalable and robust network integration. In Mashup, the diffusion in each network is first analyzed to characterize the topological context of each node. Next, the high-dimensional topological patterns in individual networks are canonically represented using low-dimensional vectors, one per gene or protein. These vectors can then be plugged into off-the-shelf machine learning methods to derive functional insights about genes or proteins. We present tools based on Mashup that achieve state-of-the-art performance in three diverse functional inference tasks: protein function prediction, gene ontology reconstruction, and genetic interaction prediction. Mashup enables deeper insights into the structure of rapidly accumulating, diverse biological network data and can be broadly applied to other network science domains.

show abstract

Section: Resultsmentioning

confidence: 92%

Section: Resultsmentioning

confidence: 99%

Section: Star Methodsmentioning

confidence: 99%

Section: Star Methodsmentioning

confidence: 99%

See 2 more Smart Citations

Compact Integration of Multi-Network Topology for Functional Analysis of Genes

Cho¹,

2016

View full text Add to dashboard Cite

show abstract

“…B. Genotypes mapped onto ontotypes that are then mapped to phenotypes. Reprinted from Yu et al (2016), Figure 1, with permission from Elsevier.…”

Section: Putting Ontologies To Work To Map From Genotypes To Phenotypesmentioning

confidence: 99%

Using the hierarchy of biological ontologies to identify mechanisms in flat networks

Bechtel

2017

Biol Philos

View full text Add to dashboard Cite

Systems biology has provided new resources for discovering and reasoning about mechanisms. In addition to generating databases of large bodies of data, systems biologists have introduced platforms such as Cytoscape to represent protein-protein interactions, gene interactions, and other data in networks. Networks are inherently flat structures. One can identify clusters of highly connected nodes, but network representations do not represent these clusters as at a higher level than their constituents. Mechanisms, however, are hierarchically organized: they can be decomposed into their parts and their activities can be decomposed into component operations. A potent bridge between flat networks and hierarchical mechanisms is provided by biological ontologies, both those curated by hand such as Gene Ontology (GO) and those extracted directly from databases such as Network Extracted Ontology (NeXO). I examine several examples in which by applying ontologies to networks, systems biologists generate new hypotheses about mechanisms and characterize these novel strategies for developing mechanistic explanations.

show abstract