Estrogen regulates divergent transcriptional and epigenetic cell states in breast cancer.

Breast cancers are known to be driven by the transcription factor estrogen receptor and its ligand estrogen. While the receptor's cis-binding elements are known to vary between tumors, heterogeneity of hormone signaling at a single-cell level is unknown. In this study, we systematically tracked estrogen response across time at a single-cell level in multiple cell line and organoid models. To accurately model these changes, we developed a computational tool (TITAN) that quantifies signaling gradients in single-cell datasets. Using this approach, we found that gene expression response to estrogen is non-uniform, with distinct cell groups expressing divergent transcriptional networks. Pathway analysis suggested the two most distinct signatures are driven separately by ER and FOXM1. We observed that FOXM1 was indeed activated by phosphorylation upon estrogen stimulation and silencing of FOXM1 attenuated the relevant gene signature. Analysis of scRNA-seq data from patient samples confirmed the existence of these divergent cell groups, with the FOXM1 signature predominantly found in ER negative cells. Further, multi-omic single-cell experiments indicated that the different cell groups have distinct chromatin accessibility states. Our results provide a comprehensive insight into ER biology at the single-cell level and potential therapeutic strategies to mitigate resistance to therapy.

Fast and interpretable genomic data analysis using multiple approximate kernel learning.

MOTIVATION: Dataset sizes in computational biology have been increased drastically with the help of improved data collection tools and increasing size of patient cohorts. Previous kernel-based machine learning algorithms proposed for increased interpretability started to fail with large sample sizes, owing to their lack of scalability. To overcome this problem, we proposed a fast and efficient multiple kernel learning (MKL) algorithm to be particularly used with large-scale data that integrates kernel approximation and group Lasso formulations into a conjoint model. Our method extracts significant and meaningful information from the genomic data while conjointly learning a model for out-of-sample prediction. It is scalable with increasing sample size by approximating instead of calculating distinct kernel matrices. RESULTS: To test our computational framework, namely, Multiple Approximate Kernel Learning (MAKL), we demonstrated our experiments on three cancer datasets and showed that MAKL is capable to outperform the baseline algorithm while using only a small fraction of the input features. We also reported selection frequencies of approximated kernel matrices associated with feature subsets (i.e. gene sets/pathways), which helps to see their relevance for the given classification task. Our fast and interpretable MKL algorithm producing sparse solutions is promising for computational biology applications considering its scalability and highly correlated structure of genomic datasets, and it can be used to discover new biomarkers and new therapeutic guidelines. AVAILABILITY AND IMPLEMENTATION: MAKL is available at https://github.com/begumbektas/makl together with the scripts that replicate the reported experiments. MAKL is also available as an R package at https://cran.r-project.org/web/packages/MAKL. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Spatiotemporal prediction of infectious diseases using structured Gaussian processes with application to Crimean-Congo hemorrhagic fever.

BACKGROUND: Infectious diseases are one of the primary healthcare problems worldwide, leading to millions of deaths annually. To develop effective control and prevention strategies, we need reliable computational tools to understand disease dynamics and to predict future cases. These computational tools can be used by policy makers to make more informed decisions. METHODOLOGY/PRINCIPAL FINDINGS: In this study, we developed a computational framework based on Gaussian processes to perform spatiotemporal prediction of infectious diseases and exploited the special structure of similarity matrices in our formulation to obtain a very efficient implementation. We then tested our framework on the problem of modeling Crimean-Congo hemorrhagic fever cases between years 2004 and 2015 in Turkey. CONCLUSIONS/SIGNIFICANCE: We showed that our Gaussian process formulation obtained better results than two frequently used standard machine learning algorithms (i.e., random forests and boosted regression trees) under temporal, spatial, and spatiotemporal prediction scenarios. These results showed that our framework has the potential to make an important contribution to public health policy makers.