DrNathanSkene

Summary

Nathan Skene is a UKRI Future Leaders Fellow and a UK Dementia Research Institute group leader at Imperial College London. His interests lie in using human genetics to gain insight into the neurobiology of brain disorders and cognitive traits. His team employs a multidisciplinary approach that includes single cell genomics, epigenomics, machine learning, and bioinformatics. 

Research interests

The lab pioneered the development of methods for genetic identification of cell types underlying complex and rare diseases. These techniques work by integrating large datasets from Genome-wide Association Studies (GWAS) and single-cell RNA-seq (scRNA-seq). They have developed the EWCE and MAGMA_Celltyping R packages which enable these analyses to be performed. Using these methods he showed that Alzheimer's disease is driven by microglia, Schizophrenia is associated with a specific subset of neurons and that Oligodendrocytes play a role in the etiology of Parkinsons.

A core focus of the lab is on identifying transcription factors which are genome-wide genetically implicated in complex traits. Towards this end, his group are developing new techniques for single cell epigenomics to resolve issues they have identified with current protocols. The methods available for epigenetic profiling have been under rapid development, and the lab has been developed statistical analysis tools for ensuring these methods behave as expected.

Machine learning is a key focus area within the lab, as these are vital for linking genetic variants to their effects. We have been using attentional deep learning models to predict the effects of genetic variants on the epigenome. We are interested in how these predictions can be validated, and how better training datasets can be created.

Reproducible research

The lab is committed to the importance of reproducible research. Towards this end we have published papers on the most appropriate statistical method for analysis of single cell RNA-seq differential expression studies. We have also published re-analyses of studies, where statistical approaches used led to false discoveries. To facilitate the take-up of reproducible research practises, we developed the Rworkflows system, which uses GitHub actions to enable easy containerisation and version control of R packages. Recognising that the ability to easily distribute and re-use GWAS summary statistics was a bottleneck for genetic research, we developed the MungeSumstats R package, which allows standardisation of these files. The scFlow workflow system was created to allow containerised and version-controlled case/control analyses of single-cell RNA-seq datasets.

Education

Nathan gained an undergraduate degree in Artificial Intelligence and Cybernetics from the University of Reading in 2008, followed by an MPhil at the University of Cambridge in Computational Biology in 2009.  He went on to do a PhD in Molecular Biology at the Wellcome Trust Sanger Institute working with Prof Seth Grant. During this time he worked on the Genes to Cognition programme, analysing the transcriptomic changes seen in a mice carrying a wide range of synaptic mutations. Later, while working between the University of Edinburgh and UCL, he studied how postnatal gene expression changes influence the onset of psychiatric disorders.

His postdoctoral research was done at the Karolinska Institutet (KI) where he worked with Jens Hjerling-Leffler as part of the Functional Neuromics project. At KI he was involved in developing large scale single cell RNA-seq atlases of brain cell types, and using these datasets to gain insight into genetic disorders. Using this approach he was able to show that multiple cell types play a role in the etiology of schizophrenia, while only microglia appear to be influenced by the common genetic factors influencing Alzheimers. 

Publications

Journals

More Publications