Differences

This shows you the differences between two versions of the page.

--- projects [2021/08/25 11:30] – mikael
+++ projects [2021/09/20 17:03] (current) – mikael
@@ Line 1: / Line 1: @@
 ==== Research projects with the Boden group ====
+Informal collection of project ideas, at different levels and duration.
 === Phylogenetics, ancestor sequence reconstruction, protein engineering, bioeconomy ===
 All biological components that are genetically encoded are subject to evolution—selective pressures in their ecological niche. With biochemists and protein engineers, we develop (phylogenetic) tools for detecting what specific changes explain the make-up of a gene or protein; this leads to a fundamental appreciation of genetic determinants of success, but can also be used to design novel, reconstruct ancient variants or even re-run evolution artificially to generate products that perform in conditions for medical, industrial and agricultural applications in the emerging bioeconomy.
-  - Inferring sequence and traits of ancestral genes and proteins
-  - Phylogenetic prospecting of proteins and their variants for protein engineering
 == Problem: What are the evolutionary drivers of metallo beta-lactamases? ==
-  * Approach: Reconstruction of the enzyme super-family incorporating MBLs and analysis of evolutionary determinants relevant to their antibiotic resistance
+  * Approach: Reconstruction of the enzyme super-family incorporating MBLs and analysis of evolutionary determinants relevant to establishing resistance to antibiotics
   * Contacts: m.boden@uq.edu.au
   * Collaborators: Hugenholtz, Schenk, Soo, Schofield
+== Problem: What ancestors in a phylogeny recapitulate a given mix of experimental properties? ==
+  * Approach: Smart interrogation of experimental databases, visualisation of phylogeny juxtaposed with available data, and novel use of evolutionary models form part of our group's broader agenda, to be incorporated in a toolkit for ancestor reconstruction.
+  * Contacts: Gabe Foley g.foley@uq.edu.au, Sam Davis sam.davis@uq.edu.au, m.boden@uq.edu.au
+----
 === Analytical tools for sequencing data, omics data integration methods, machine learning ===
-== What are the genetic drivers in development and disease? ==
+New (sequencing) technologies bring new opportunities and challenges, including long- and short read technologies at single-cell or bulk resolution, and with spatial specificity. We develop tools for leveraging the scale of available data, including assessment of reproducibility, the discovery and extraction of biological footprints that emerge across time, in 3D and different cellular conditions, or that are available by complementarity between data types, including sequence motifs. We regularly publish prediction services and computational tools open to the scientific community.
-Bioinformatics methods for understanding transcriptional and epigenetic regulation during mammalian development and in disease is a significant focus. We are relying on technologies such as single-cell RNA-seq and ChIP-seq, in collaboration with developmental, neuroscience and cell biologists at UQ and elsewhere.
+== Problem: Assessing reproducibility of ChIP-seq does not address intra-experimental bias ==
-  - Data integration for understanding genetic drivers in development and disease
+  * Approach: The group previously developed ChIP-R to evaluate reproducibility of ChIP-seq replicates, by adapting the rank-product test, applied to each site independently; this approach does not consider the within-experiment dependence, exemplified by a quality metric (or even a profile) for each replicate. This project should investigate the basis for a *model* (of each experimental replicate), and evaluate an implementation.
-  - Statistically analysing and extracting biological meaning from genome-wide data
+  * Contact: m.boden@uq.edu.au
-== Problem: What ancestors should be prioritised to identify those with optimal properties? ==
+== Problem: de-convoluting bulk data as a mixture of cell types: a case of using variance in replicates? ==
-  * Approach: Experimental data is typically scarce and difficult to juxtapose on phylogenies.
-  Models of evolution to triangulate successful ancestral catalysts
+  * Approach: multiple biological replicates may represent different mixtures of cell types, hence variance across replicates may reveal what signals that originate from one type and not the others. Evaluate if (for instance) latent, cell type-mixture models could find coefficients that would allow de-mixing, i.e. separating the source behind each signal? This would be particularly helpful on ChIP-seq data, which is currently not at single-cell resolution, but similar ATAC-seq data can be used to benchmark by using matched single-cell data.
-  * Contacts:
+  * Contact: m.boden@uq.edu.au
-  * Collaborators:
+== Problem: how can we link cell states identified in one single-cell RNA-seq experiment to another? ==
+  * Approach: identify a universal representation of cell states that recapitulate statistical properties in gene expression, say by data matrix decomposition or similar with machine learning
+  * Contact: a.mora@uq.edu.au, b.balderson@uq.edu.au, m.boden@uq.edu.au
+  * Collaborator: Thor
+== Problem: Interrogation of single-cell ATAC-seq and integration with single-cell RNA-seq ==
+  * Approach: Regulatory markers are present but look different in ATAC- and RNA-seq. We expect the mapping of representative cells in each data type could be based on regulatory footprints; this needs to be explored via the use of appropriate models that accommodate the uncertainty of inter-set labelling.
+  * Contact: m.boden@uq.edu.au, a.mora@uq.edu.au
+  * Collaborator: Palpant
+----
 === Models of development and disease across time and space ===
-  * Problem:
+Increasingly genome technologies uncover spatial and temporal specificity of observations, but data need to be carefully and selectively pieced together. We work on integrating genomic, transcriptomic and epigenomic data, viewed together with information that can be predicted from sequence and other genomic markers, accounting for the uncertainty of their juxtaposition. With collaborators we have been using data integration to infer drivers in development (of the brain and other organs) as well as in cancer and disease.
-  * Approach:
-  * Contacts:
+== Problem: To what extent is epigenetic regulation conserved, and how can evolution help interpret and explain epigenetic regulation? ==
-  * Collaborators:
+  * Approach: The conservation of histone-modifying enzymes is broadly appreciated to exist in many higher species, and is especially relevant to the development of organs. The plan is to probe and model variance //within// species, trace conservation and model evolution //across// species via (on the one hand) positioning of epigenetic marks (at key developmental stages) and sequence and expression of epigenetic components.
+  * Contact: m.boden@uq.edu.au
+  * Collaborator: Thor
+== Problem: How can we leverage trends of epigenetic marks to highlight regulatory drivers in sparse single-cell and spatial transcriptomes? ==
+  * Approach: Evaluate how key histone modifications (collected by ENCODE, say) act coordinately with gene expression; model such coordination with view of predicting epigenetic regulation
+  * Contact: m.boden@uq.edu.au
+== Problem: What is the quantitative nature of DNA methylation and its role in cancer? ==
+  * Approach: Data aggregation and careful integration of massive public data sets give rise to new hypotheses.
+  * Contact: Ariane Mora a.mora@uq.edu.au, m.boden@uq.edu.au