I thought it may be useful to start this thread to brainstorm what computational tools (mathematical modeling / bioinformatics) we have as a group that could be extended through collaboration on distributed among projects?
To introduce myself, I am primarily on the bioinformatics side. I focus on analysis of gene expression, DNA methylation, and more recently ChIP-seq data of chromatin structure. My primary computational tools are a Bayesian non-negative matrix factorization algorithm, CoGAPS (https://academic.oup.com/bioinformatics/article-abstract/doi/10.1093/bioinformatics/btx058/2975325/PatternMarkers-amp-GWCoGAPS-for-novel-data-driven?redirectedFrom=fulltext) which is useful for determining gene patterns associated with inferred biological processes. Our approach can also separate genes that are biomarkers from ones that are reused in multiple biological processes and has been extended to data integration of DNA methylation / gene expression (http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0078127).
Recently, we have also been developing techniques to account for inter-tumor heterogeneity in pathway dysregulation (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4218688/) or gene isoform usage (http://biorxiv.org/content/early/2016/12/05/091637).
I am very interested in thinking about extensions to these algorithms to other applications and also how they may be integrated with mathematical models -- potentially using the results of mathematical models as covariates in the bioinformatics analysis or using the bioinformatics tools to inform parameters of the mathematical model from genomics data.
Created by Elana Fertig elana.fertig Bioinformatics for cell motility and cell fate would also be useful, both from a single cell-fate analysis and populations level analysis (i.e. stem cell lineage tracing over time).
@JAguirre-Ghiso thanks for that advice.
These tools are pretty diverse and can be used for a lot of applications.
Some examples that are in the literature are / we are currently working on:
(1) finding tumor subtypes that have shared DNA methylation and gene expression changes http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0078127
(2) determining gene signatures that are associated with different pathway perturbations and their relationship to therapeutic sensitivity: http://www.impactjournals.com/oncotarget/index.php?journal=oncotarget&page=article&op=view&path%5B%5D=12075
(3) determining the timing of changes in gene expression or DNA methylation in different experimental conditions.
Some applications that I can see off the top of my head to these projects would be:
(1) gene expression or DNA methylation relationships among primary tumors / metastatic samples / and healthy organs to which a tumor metastasizes
(2) what genes change in the transition from invasion to metastasis
(3) integrating these algorithms with mathematical models of invasion to predict phenotypic / genomic changes associated with metastasis
Ideally, the tool could be tailored to the biological hypothesis being tested that the group comes up with.
It would be great to get feedback from the other computational folks in the group as well. For the non-math crowd.
@lanilonzo
Some examples as to what data you input and what predictions you can make with the different tool would be useful. Maybe some discrete simplified examples.
Thanks