Knowledge-Based Analysis for Detecting Key Signaling Events from Time-Series Phosphoproteomics Data

Humphrey

Cinghu

et al. 2018

Preprint

Self Cite

Pluripotency is highly dynamic and progresses through a continuum of pluripotent stem-cell states. The two states that bookend the pluripotency continuum, naïve and primed, are well characterized, but our understanding of the intermediate states and transitions between them remain incomplete. Here, we dissect the dynamics of pluripotent state transitions underlying pre-to post-implantation epiblast differentiation. Through integrative analysis of the proteome, phosphoproteome, transcriptome, and epigenome of embryonic stem cells transitioning from naïve to primed pluripotency, we show that rapid, acute, and widespread changes to the phosphoproteome precede ordered changes to the epigenome, transcriptome, and proteome. Reconstruction of kinase-substrate networks reveals signaling cascades, dynamics, and crosstalk. Distinct waves of global proteomic changes mark discrete phases of pluripotency, characterized by cell surface markers that track pluripotent state transitions. Our data provide new insights into the multi-layered control of the phased progression of pluripotency and a foundation for modeling mechanisms regulating pluripotent state transitions. HIGHLIGHTS • Multi-ome maps of cells transitioning from naïve to primed pluripotency • Phosphoproteome dynamics precede changes to epigenome, transcriptome, and proteome • Kinase-substrate network reconstruction uncovers signaling dynamics and crosstalk • Proteins and cell surface markers that track pluripotent state transitions • Comparative analysis of mouse and human pluripotent states

“…Yang, P., Zheng, X., Jayaswal, V., Hu, G., Yang, J.Y., and Jothi, R. (2015). Knowledge-Based Analysis for Detecting Key Signaling Events from Time-Series Phosphoproteomics Data.…”

Section: Conflict Of Interestmentioning

confidence: 99%

Multi-Omic Profiling Reveals Dynamics of the Phased Progression of Pluripotency

Humphrey

Cinghu

et al. 2018

Preprint

Self Cite

“…These significantly changed profiles were then standardized (see above) and partitioned into 17 clusters using the FCM 3 algorithm in the Mfuzz R package 22 ( Supplementary Table 1). The partitioning into 17 clusters was reported to be optimal by Yang et al 6 for this data set.…”

Section: Phosphoproteomics Data Setmentioning

confidence: 87%

“…We also tested Minardo-Model by applying it to a multiomics data set published by Yang et al 8 , which measured responses in mouse E14Tg2a cells as they differentiated from the naive embryonic stem cell (ESC) state to primed epiblast-like cells (EpiLC) 8 . Yang et al measured phosphoproteomics changes at 12 time points (including basal) and provided 3,585 phosphosite profiles partitioned into 4 clusters using CLUE 6 . We removed 21 profiles, since their corresponding protein name was not provided, and the remaining 3,564 profiles were analysed with Minardo-Model ( Supplementary Fig.…”

Section: Multiomics Data Setmentioning

confidence: 99%

“…The inferred events are then overlaid onto the widely-used matrix of profile plots ( Fig. 1, Supplementary Figs 4-6), but where individual profile variation has been colored in a manner which avoids limitations of rainbow color maps 2,6 used in some previous work 2, 6 . The novel event map and event sparkline layouts are useful for revealing overall trends, while still providing sufficient detail (via cluster numbering and event ordering) to provide insight into the behavior of individual biomolecules.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Temporal ordering of omics and multiomic events inferred from time series data

Kaur

Luu

et al. 2020

Preprint

Self Cite

Temporal changes in omics events can now be routinely measured, however current analysis methods are often inadequate, especially for multiomics experiments. We report a novel analysis method that can infer event ordering at better temporal resolution than the experiment, and integrates omic events into two concise visualizations (event maps and sparklines). Testing our method gave results well-correlated with prior knowledge and indicated it streamlines analysis of time-series data. MainA range of emerging omics and multiomics techniques now provide unprecedented ability to systematically track the time course of changes in abundance, for example, in mRNAs, proteins, or posttranslational modifications (PTMs). The resulting time-series data can then be used to infer the ordering of underlying events (e.g., activation of kinases, transcription factors, or translation factors) that explain the observed changes in abundance. These ordered events provide insight into the causal flow of information underlying transitions in cellular state. These insights, in turn, have the potential to lead to fundamental advances in biology and biomedical sciences -however, realizing this potential requires improving the analysis methods used, which are often inadequate with existing data sets 1,2 , especially from multiomics experiments.When analysing data from such experiments, a common first step is to partition the (typically) tens of thousands of abundance time profiles found into tens of clusters, each with similar temporal changes that are then visualized in a single profile plot 1,2 (Fig. 1). A wide variety of clustering methods are available, ranging from widely-used, general purpose methods (e.g., fuzzy c-means; FCM 3 ) to specialized methods tailored for specific, omics experimental scenarios 4 , 5 , 6 . The specialist methods typically combine clustering with assigning the underlying events, for example, to specific kinases or transcription factors. Often, these events are further assigned to specific time intervals used in the experiment, or to approximate time windows (e.g., early, intermediate, or late response), thus defining an implicit event ordering. However, a key limitation of current approaches is seen when plotting the inferred events using temporal-based layouts 7 : there can often be many more events than time intervals in the original experiment,

“…This resulted in 3178 regulated phosphorylation sites. We then applied knowledge-based cluster evaluation (CLUE) (Yang et al, 2015) to determine the optimal clustering of the phosphorylation sites and identified kinases whose substrates are enriched within each cluster. Using CLUE, we identified four clusters enriched with known substrates of a set of kinases (Fig.…”

Section: Kinase Selection and Feature Extractionmentioning

confidence: 99%

Positive-unlabeled ensemble learning for kinase substrate prediction from dynamic phosphoproteomics data

Humphrey

James³

et al. 2015

Bioinformatics

Self Cite

Motivation: Protein phosphorylation is a post-translational modification that underlines various aspects of cellular signaling. A key step to reconstructing signaling networks involves identification of the set of all kinases and their substrates. Experimental characterization of kinase substrates is both expensive and time-consuming. To expedite the discovery of novel substrates, computational approaches based on kinase recognition sequence (motifs) from known substrates, protein structure, interaction and co-localization have been proposed. However, rarely do these methods take into account the dynamic responses of signaling cascades measured from in vivo cellular systems. Given that recent advances in mass spectrometry-based technologies make it possible to quantify phosphorylation on a proteome-wide scale, computational approaches that can integrate static features with dynamic phosphoproteome data would greatly facilitate the prediction of biologically relevant kinase-specific substrates. Results: Here, we propose a positive-unlabeled ensemble learning approach that integrates dynamic phosphoproteomics data with static kinase recognition motifs to predict novel substrates for kinases of interest. We extended a positive-unlabeled learning technique for an ensemble model, which significantly improves prediction sensitivity on novel substrates of kinases while retaining high specificity. We evaluated the performance of the proposed model using simulation studies and subsequently applied it to predict novel substrates of key kinases relevant to insulin signaling. Our analyses show that static sequence motifs and dynamic phosphoproteomics data are complementary and that the proposed integrated model performs better than methods relying only on static information for accurate prediction of kinase-specific substrates. Availability and implementation: Executable GUI tool, source code and documentation are freely available at https://github.