Computational Analysis of Transcription and Alternative Splicing Events in Squamous Cell Cancer.

Squamous cell carcinomas (SCCs) are a very common type of human cancer which affect surface in the body, such as the skin, cervix (neck of the womb) and the lining of the mouth and throat. There has been little progress towards finding new and successful treatments for these important cancers, although recent studies have drugs which target sets of genes that drive the SCC to occur do help. I will study in detail how SCCs are affected by a process called "alternative RNA splicing".

The DNA in our cells contains a code which is 'read' to make messenger molecules called RNA. The useful bit of the code which is turned into RNA are the genes, and within each gene, the 'words' of the message are alternated with areas of irrelevant code. To get a meaningful message, the useful words need to be 'spliced' together and the irrelevant bits spliced out. That process is called RNA splicing. In some genes, more than one meaningful message can be made from one set of DNA words by 'alternative RNA splicing'. That can be part of how cells function normally, but if it goes wrong it can damage the message produced, potentially making a cell function less well or to grow out of control (as in cancer).

Controlling this splicing involves a complex mix of proteins, including 'splicing factors' and 'RNA binding proteins'. RNA splicing is changed in cancer cells but for individual cancer types, we do not understand the full range of the changes, nor which of them would make the best targets for new treatments or tests.

I will identify splicing changes that are common to SCCs, and also those that are specific to a particular site and/or cause. I will look at samples from patients' SCCs that come from two different sites in the body - the skin and the cervix. These have different causes, particularly sunlight in the skin and infection with a virus (the human papillomavirus) in the cervix.

I will use powerful new technologies to read the sequence of all the RNA molecules in the samples. I will join a unique collaboration between the European Bioinformatics Institute (EBI) and Cambridge University. I will be based with Dr Anton Enright at the EBI, a world-renowned centre for biological computing near Cambridge, and will undertake the experimental and clinical aspects of the project in the laboratory of Professor Nicholas Coleman at Cambridge University Department of Pathology. Together, these highly complementary environments will provide me with the ideal platform to complete this work successfully and to develop my career as an academic doctor.

Using databases of gene studies from large projects in America, I will be able to see if the changes I've found make a difference to how long patients with cancer survive and respond to drugs. These databases have information for a wide range of cancers, including various SCCs, but none for normal, healthy tissue - which is why this project is needed.

All the results I generate will be made available openly using a secure database that can be accessed through the internet, and any scientific papers that I publish based on this work will be freely available to anyone to read.

My project will greatly improve our understanding of how RNA splicing works in SCC cancer cells and will identify differences between SCCs that have particular causes and/or occur at specific sites in the body. The work will open up new opportunities for diagnosing and/or treating a variety of common SCCs.

This project focuses on comprehensively measuring alternative RNA splicing in squamous cell carcinoma (SCC). I will compare SCCs from two different anatomical sites, which arise through distinct aetio-pathological processes:

(i) Skin SCC, which is primarily due to ultraviolet radiation exposure
(ii) Cervical SCC, caused by high-risk human papillomavirus (HRHPV) infection.

From 40 tumour and 20 site, age and gender matched samples from healthy (no cancer) individuals) I will generate high-throughput strand-specific RNA-sequencing data and build transcriptome assemblies. I will compare RNA-seq tools (eg. TopHat/Bowtie/Cufflinks, BWA, Scripture, Kallisto, Sailfish, Salmon, Trinity) and assess their accuracy using RASE-qPCR to determine the optimum toolchain for RNA-seq splicing analysis.

I will quantify isoform expression in each tissue sample and quantify alternative splicing events (eg. SUPPA, MISO, MATS).

I will investigate the expression of different splicing factors and their targets, as well as differential isoform expression of important cancer genes. Tools including DESeq, DEXSeq, Sleuth, edgeR, voom will be compared and optimised.

Expression will be correlated to clinico-pathological features (age, gender, tumour stage and grade) for each cancer type using a support vector machine classifier and will be tested on independent sample sets. Biologically relevant candidates will be validated experimentally by qRT-PCR and gene over-expression/knockout. I will identify targets bound by cancer-specific splicing factors using cross-linking immunoprecipitation high-throughput sequencing.

This work will improve our understanding of the role of alternative splicing in SCCs, the mechanisms of splicing deregulation and the functional consequences of altered splicing for SCC-specific regulatory networks. It will also indicate candidate diagnostic and/or prognostic biomarkers and potential targets for novel therapies in SCC

Through this training fellowship I will become a highly skilled clinical academic with expertise in both the computational biology field, and the fundamental biology of cancer pathogenesis. The project will provide training in a wide range of computational and analytic techniques, complimented by formal teaching sessions at EMBL-EBI and at the University of Cambridge School of Graduate Studies. There are very few clinicians with strong computational biology skills and the ability to bridge the divides between the clinical, research and bioinformatics fields. By the completion of this work I will be well placed to lead the clinical academic field in this regard. As clinical and research genomics become standard it is vital that that there are clinicians with a genuine understanding of the methodologies and tools required for analysis and interpretation of this data.

The primary sequencing dataset will be of significant value to large numbers of groups studying gene expression in cancer, and parties interested in larger scale studies across cancer will be able to integrate these datasets into large meta-studies. Consequently there will be significant impact for both fundamental and clinical scientists.

The work proposed will improve our understanding of alternative splicing and how it contributes to organism complexity. Researchers involved in splicing, transcriptional regulation and post-transcriptional control will will be impacted by these studies. The use of high-throughput CLIP assays will be useful to the growing community of scientists involved in the application and analysis of RNA bound to protein. The research will generate novel computational pipelines and methods for the computational analysis and integration of high-throughput data. These tools will be of use to computational biologists and genomics scientists and may be reused for other studies and applications.

Clinical research scientists and particularly those involved in cancer will find the results useful and applicable to other cancers. As stated in the data-management plan and pathways to impact documents all data and downstream analysis results will be deposited in public repositories and made available through visualisation on the world-wide-web and via tracks for popular genome-browsers such as IGV, UCSC or Ensembl. The results generated will contribute to the growing wealth of post-genomics knowledge being generated around the world.

There are impacts for the knowledge economy of the United Kingdom as this cutting edge-research brings together experts in computational genomics and clinical sciences to tackle a common problem. The UK must take a lead on new high-throughput genomics technologies in order to retain its competitiveness in the sciences. There are additional economic and commercial impacts. In particular, there is genuine potential for drug targets or novel diagnostic/prognostic tools to be exploited by industrial partners.

The goal of this project is to derive new insights into cancer and tumorigenesis with the potential for diagnostic or therapeutic targets. This impact long term will be felt by patients through improvements in diagnosis, prognostic tools and ultimately treatment of common and serious cancers.