MDI-GPU: accelerating integrative modelling for genomic-scale data using GP-GPU computing

Mason, Samuel A; Sayyid, Faiz; Kirk, Paul; Starr, Colin; Wild, David L.

doi:10.1515/sagmb-2015-0055

Cited by 8 publications

(6 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One approach (Zhang et al, 2013) uses super k-means clustering of an integrated data matrix to identify modules. Another approach, MDI (Kirk et al, 2012; Mason et al, 2016) uses a Bayesian model to capture relationships between gene sets. When the three methods were run using the most variable features from each dataset, we found that MDI and super k-means do not enrich for PPIs at the same level as MAGNETIC for raw networks as well as networks not used in the construction of MAGNETIC (Figure S3a-e, see implementation of related approaches section).…”

Section: Star Methodsmentioning

confidence: 99%

Integration of Tumor Genomic Data with Cell Lines Using Multi-dimensional Network Modules Improves Cancer Pharmacogenomics

2018

View full text Add to dashboard Cite

Leveraging insights from genomic studies of patient tumors is limited by the discordance between these tumors and the cell line models used for functional studies. We integrate -omics datasets using functional networks to identify gene modules reflecting variation between tumors and show that the structure of these modules can be evaluated in cell lines to discover clinically relevant biomarkers of therapeutic responses. Applied to breast cancer, we identify 219 gene modules that capture recurrent alterations, subtype patients and quantitate various cell types within the tumor microenvironment. Comparison of modules between tumors and cell lines reveals that many modules composed primarily of gene expression and methylation are poorly preserved. In contrast, preserved modules are highly predictive of drug responses in a manner that is robust and clinically relevant. This work addresses a fundamental challenge in pharmacogenomics that can only be overcome by the joint analysis of patient and cell line data.

show abstract

Section: Star Methodsmentioning

confidence: 99%

Integration of Tumor Genomic Data with Cell Lines Using Multi-dimensional Network Modules Improves Cancer Pharmacogenomics

2018

View full text Add to dashboard Cite

show abstract

“…Bayesian inference was performed using MCMC (see ref. 55 for further details on inference and modelling) as implemented in the MDI-GPU software 98 . Clusters were extracted only for proteins that were consistently allocated to the same clusterings (sampled from the posterior distribution) across replicates 55,99 (Supplementary Data 3).…”

Section: Marker List Generationmentioning

confidence: 99%

Spatiotemporal proteomic profiling of the pro-inflammatory response to lipopolysaccharide in the THP-1 human leukaemia cell line

et al. 2021

View full text Add to dashboard Cite

Protein localisation and translocation between intracellular compartments underlie almost all physiological processes. The hyperLOPIT proteomics platform combines mass spectrometry with state-of-the-art machine learning to map the subcellular location of thousands of proteins simultaneously. We combine global proteome analysis with hyperLOPIT in a fully Bayesian framework to elucidate spatiotemporal proteomic changes during a lipopolysaccharide (LPS)-induced inflammatory response. We report a highly dynamic proteome in terms of both protein abundance and subcellular localisation, with alterations in the interferon response, endo-lysosomal system, plasma membrane reorganisation and cell migration. Proteins not previously associated with an LPS response were found to relocalise upon stimulation, the functional consequences of which are still unclear. By quantifying proteome-wide uncertainty through Bayesian modelling, a necessary role for protein relocalisation and the importance of taking a holistic overview of the LPS-driven immune response has been revealed. The data are showcased as an interactive application freely available for the scientific community.

show abstract

“…We use the implementation of Bayesian mixture models in C++ provided by Mason et al (2016). Rather than directly using a Dirichlet process (Ferguson, 1973) to infer the number of clusters or a mixture that grows and shrinks (Richardson and Green, 1997), this implementation follows the logic of Rousseau and Mengersen (2011) and Van Havre et al (2015) using an overfitted mixture model to approximate a Dirichlet process.…”

Section: Supplementary Materialsmentioning

confidence: 99%

Consensus clustering for Bayesian mixture models

Coleman

Kirk

Wallace

2020

Preprint

Self Cite

View full text Add to dashboard Cite

MotivationCluster analysis is an integral part of precision medicine and systems biology, used to define groups of patients or biomolecules. However, problems such as choosing the number of clusters and issues with high dimensional data arise consistently. An ensemble approach, such as consensus clustering, can overcome some of the difficulties associated with high dimensional data, frequently exploring more relevant clustering solutions than individual models. Another tool for cluster analysis, Bayesian mixture modelling, has alternative advantages, including the ability to infer the number of clusters present and extensibility. However, inference of these models is often performed using Markov-chain Monte Carlo (MCMC) methods which can suffer from problems such as poor exploration of the posterior distribution and long runtimes. This makes applying Bayesian mixture models and their extensions to ‘omics data challenging. We apply consensus clustering to Bayesian mixture models to address these problems.ResultsConsensus clustering of Bayesian mixture models successfully finds generating structure in our simulation study and captures multiple modes in the likelihood surface. This approach also offers significant reductions in runtime compared to traditional Bayesian inference when a parallel environment is available. We propose a heuristic to decide upon ensemble size and then apply consensus clustering to Multiple Dataset Integration, an extension of Bayesian mixture models for integrative analyses, on three ‘omics datasets for budding yeast. We find clusters of genes that are co-expressed and have common regulatory proteins which we validate using external knowledge, showing consensus clustering can be applied to any MCMC-based clustering method.

show abstract

MDI-GPU: accelerating integrative modelling for genomic-scale data using GP-GPU computing

Cited by 8 publications

References 9 publications

Integration of Tumor Genomic Data with Cell Lines Using Multi-dimensional Network Modules Improves Cancer Pharmacogenomics

Integration of Tumor Genomic Data with Cell Lines Using Multi-dimensional Network Modules Improves Cancer Pharmacogenomics

Spatiotemporal proteomic profiling of the pro-inflammatory response to lipopolysaccharide in the THP-1 human leukaemia cell line

Consensus clustering for Bayesian mixture models

Contact Info

Product

Resources

About