Wrangling Phosphoproteomic Data to Elucidate Cancer Signaling Pathways

Abstract

The interpretation of biological data sets is essential for generating hypotheses that guide research, yet modern methods of global analysis challenge our ability to discern meaningful patterns and then convey results in a way that can be easily appreciated. Proteomic data is especially challenging because mass spectrometry detectors often miss peptides in complex samples, resulting in sparsely populated data sets. Using the R programming language and techniques from the field of pattern recognition, we have devised methods to resolve and evaluate clusters of proteins related by their pattern of expression in different samples in proteomic data sets. We examined tyrosine phosphoproteomic data from lung cancer samples. We calculated dissimilarities between the proteins based on Pearson or Spearman correlations and on Euclidean distances, whilst dealing with large amounts of missing data. The dissimilarities were then used as feature vectors in clustering and visualization algorithms. The quality of the clusterings and visualizations were evaluated internally based on the primary data and externally based on gene ontology and protein interaction networks. The results show that t-distributed stochastic neighbor embedding (t-SNE) followed by minimum spanning tree methods groups sparse proteomic data into meaningful clusters more effectively than other methods such as k-means and classical multidimensional scaling. Furthermore, our results show that using a combination of Spearman correlation and Euclidean distance as a dissimilarity representation increases the resolution of clusters. Our analyses show that many clusters contain one or more tyrosine kinases and include known effectors as well as proteins with no known interactions. Visualizing these clusters as networks elucidated previously unknown tyrosine kinase signal transduction pathways that drive cancer. Our approach can be applied to other data types, and can be easily adopted because open source software packages are employed.

FlyClockbase: Importance of Biological Model Curation for Analyzing Variability in the Circadian Clock of Drosophila melanogaster by Integrating Time Series from 25 Years of Research

“…Problems with tracing the identity, availability, accuracy, precision, and reliability of data have been the topic of numerous investigations in various contexts, some of which involve big data (e.g. see (R eason and M ycielska 1982; R eason 1990; R eason and H obbs 2003; G oldston 2008; D oan et al 2012; G itelman 2013; G rimes et al 2013; M c C allum 2013; R eason 2013; B lankenberg et al 2014; R eason 2015)).…”

“…While designers of biological information resources can learn much from the decades of research that informed the development of human error analysis tools in those areas, it is less straight forward how these insights could be applied to improve the quality of biological information available to most biologists. A source of concern is the substantial complexity of many human error analysis frameworks (R eason and M ycielska 1982; R eason 1990; NASA et al 2001-09-30; R eason and H obbs 2003; NASA et al 2006-07; G oldston 2008; NASA et al 2011; G itelman 2013; G rimes et al 2013; M c C allum 2013; R eason 2013; B lankenberg et al 2014; R eason 2015; P anko 2016). Most of these frameworks will handle the complexity of biological data, but require near prohibitive research and implementation efforts that make integration into grass roots VBIR projects such as FlyClockbase not efficient if started by biologists.…”

“…see (R eason and M ycielska 1982; R eason 1990; R eason and H obbs 2003; G oldston 2008; D oan et al . 2012; G itelman 2013; G rimes et al . 2013; M c C allum 2013; R eason 2013; B lankenberg et al .…”

RCytoscape: tools for exploratory network analysis

“…Recently, Grimes et al, [26] used RCytoscape to graph nodes in two and three dimensions using coordinates derived from dimension reduction (embedding) algorithms, and plot networks using protein-protein interaction edges merged from three different databases, setting visual properties for nodes and edges based on quantitative information from a lung cancer phosphoproteomic data set. RCytoscape will be useful to anyone who wishes to add network visualization and analysis to the rich resources available in the R programming environment.…”

Phosphoproteomics Analysis Identifies Novel Candidate Substrates of the Nonreceptor Tyrosine Kinase, Src-related Kinase Lacking C-terminal Regulatory Tyrosine and N-terminal Myristoylation Sites (SRMS)

“…A threshold Andromeda score of 40 and a threshold delta score of 8 was applied to the identified phosphopeptides, in accordance with parameters described previously (33)(34)(35). Pearsons's correlation coefficients were determined in R using log 2 transformed phosphopeptide intensities, as described before (36,37). The MaxQuant file output designated "Phospho(STY)sites" used for further analysis using the Perseus software (Version 1.5.0.15).…”