Research banner  

Overview


 

Much of our research is at the intersection of virology and immunology. Currently our primary focus is on intelligent vaccine design, but we have several other active areas of research including: predicting adverse responses to therapeutic proteins; the identification of effective cancer biomarkers; and biomedical text mining (e.g. automated metabolic pathway construction). All of our work is computational, but we work in close collaboration with experimental groups and clinicians.

 

Immune repertoires and vaccine design


 

Many research groups now believe that, in order to effectively tackle rapidly-mutating viruses such HIV and influenza A, it will be necessary to design multiple immunogens capable of guiding the host immune response down a particular maturation pathway that culminates in the induction of broadly neutralising antibodies. Such approaches require a deep understanding of host immune responses. To this end, we analyse antibody repertoire data from Next Generation Sequencing (NGS), tandem mass spectrometry (MS/MS) and hybridoma experiments in collaboration with Prof Michael Cho's lab (Iowa State). We are currently working on a range of innovative approaches to the analysis of "antibodyomes" in the following areas: benchmarking, simulation, pathway analysis, and lineage visualization.

Group members: Dr William Lees (post-doc), Rao Gudivada and James Taylor (PhD students)
Former group members: Charlie Bird, Anna Laddach and Zheli Tan (MRes students)

 

Analysis of an antibody repertoire after vaccination. Left: Network diagram showing the abundance of a small subset of heavy chain CDR3s in the vicinity of a neutralising antibody of interest. Sequences identified using the AbMining ToolBox (D'Angelo et al., 2014). Centre: Length distribution of heavy chain CDR3s. Right: Lineage plot showing light chain CDR3s after sequence assembly.

 

Conservation and antigenic escape


 

We have developed an "epitope clustering" method for predicting antigenic escape from antibodies by rapidly mutating viruses such as influenza A (Lees et al., 2011). We have used this approach to challenge assumptions about vaccine targets in the stalk of the key surface glycoprotein haemagglutinin (our paper on this topic [Lees et al., 2014] featured here on the blog of the Society for General Microbiology) and assumptions about the shielding role of N-glycans (Pentiah et al., 2014). We are currently refining our approach drawing on a long-standing interest in the properties of epitopes (Sivalingham & Shepherd, 2010) and a growing interest in molecular mimicry. We are also dissecting patterns of antibody cross-reactivity between flu subtypes in collaboration with Nigel Temperton's lab.

Group members: Dr William Lees (post-doc)
Former group members: , Kevin Pentiah (PhD student), Paddy McMahon (MSc student), Ganesh Sivalingham (PhD rotation project student), Claire Winship (honorary research associate), Rute Magalhaes (MSc student)

 
   
 
    

Epitopes and antigenic escape. Top left: Distribution of key energetic residues on three structural epitopes (PDBs 1NBZ, 1VBF and 1JRH). Bottom left: Amino acid frequency plot for a single haemagglutinin residue showing antigenic activity in the period 1988-97 prior to N-glycosylation. Right: Canonical antigenic sites of influenza A H3N2 haemagglutinin.

 

T cell immunology: therapeutic proteins, tumour antigen and viruses


 

The Shepherd Group has several collaborations in which T cell immunity is the central focus.

Haemophilia A (HA) is an example of a disease caused by a mutated protein, in this case factor VIII, part of the coagulation cascade. HA is commonly treated by replacement therapy, but the immune systems of a subset of patients attack the replacement factor VIII (rFVIII). In collaboration with Dr Dan Hart (Barts) and Dr Keith Gomez (Royal Free), the key question we are addressing is: given knowledge of a patient's disease-causing mutation, their HLA alleles (i.e. what set of MHC molecules they have) and the form of rFVIII they are treated with, can we predict whether they are at risk of an immune response to rFVIII? (Shepherd et al., 2014).

In collaboration with Dr Shilpa Chokshi (Institute of Hepatology), we are working on the selection of effective biomarkers for hepatocellular carcinoma and dissecting the T cell responses to hepatitis B virus. We are also collaborating with Prof Oliver Garden (RVC) to gain insights in/d/as9/d/shepherd-group/htdoc/to the evolution of tolerance regulation, with potential relevance to the therapeutic manipulation of regulatory T cells (Tregs) for a range of medical conditions.

We also have a broad interest in understanding the patterns of T cell epitope immunodominance in viruses.

Group members: Stuart Skelton (RA and PhD student), Naz Uzun (PhD student), Michael Denyer (PhD student and former MRes student)
Former group members: Ginny Devonshire, Paula Luz Horne, Alison Kakoschke, Devlatha Nanga and Dr Alessandra Berto (MSc students), Claire Haskins (MRes students)

 
heatmap  

Haemophilia and hepatitis B virus. Left: Heatmap showing the predicted risk of antibody development associated with different combinations of HLA-DR allele (y-axis) and Factor VIII mutation (x-axis). Right: The frequency (y-axis) of epitopes in the hepatitus B virus core antigen with different binding affinities (x-axis) for different class II HLA alleles that are known to be protective.

 

Biomedical text mining


 

The group has a track record in the development and evaluation of methods for extracting the relationships between biomedical entities (e.g. genes, proteins, metabolites) from journal articles (Clegg & Shepherd 2007; Kabiljo et al., 2009; Czarncki et al., 2012). Our current focus is on the automated construction and annotation of metabolic pathways from text articles in a collaboration with Unilever.

Former group members: Jan Czarnecki, Andrew Clegg and Renata Kabiljo (PhD students), Sander van Boom (intern)