IMMUNOLOGY EDITORIAL

Bioinformatics for immunologists

Daniel M. Altmann Summary

Department of Medicine, Hammersmith
Hospital, Du Cane Road, London, W12
Immunology was once a specialty prone to cause dismay or even scepti-
0NN, UK cism among outsiders for its struggles to visualize poorly understood,
complex interactions through descriptive models integrating cell types,
their factors and functions. This was the age of ‘too many soft ideas
propped up by too little hard data’. Twenty-first century immunologists
have the advantage of being able to marry this rich conceptual legacy to a
contemporary toolkit offering such depth of hard data across different
‘omics’ platforms, that they are faced by the opposite dilemma: ‘too much
hard data to comprehend or synthesize into a meaningful narrative’.
Approaches including next-generation sequencing of host and pathogen
genomes and transcriptomes, metagenomics of the microbiota, creative
strategies for receptor repertoire sequencing, and then for proteomics and
metabolomics, encompass all that is needed to tell the entire story, if only
we are creative enough, not only to evaluate the message from any given
omics platform, but to derive the tools that enable us to integrate the
doi:10.1111/imm.12987
Correspondence: Department of Medicine,
answers from diverse omics platforms in a meaningful way. To achieve
Hammersmith Hospital, Du Cane Road, this goal, there is an urgent need to ensure we train the next generation
London, W12 0NN, UK of bioinformatically literate researchers.
Email: d.altmann@ic.ac.uk

A graduate student reading papers on T-cell receptor set to other, related sets. In the field of infection and
(TCR) or immunoglobulin sequence data and interpreta- immunity, a classic conundrum is that we tend to have
tion from the 1990s might be surprised to find that many separated silos of pathogen sequencers and sequencers of
of the key answers were elucidated from sequenced panels host responses to pathogens, but few who have been able
that aligned repertoires of perhaps 50 or fewer sequences. to grasp the nettle and relate the two data sets to each
The answers were often surprisingly illuminating, yet any- other so as to narrate both sides of the arms race simulta-
one could spot the dangers inherent in drawing conclu- neously. The emerging methodologies for ‘systems serol-
sions about a repertoire of around 100 million expressed ogy’ testify to the added value of integrating data sets
receptors, based on a sample of 50. This was akin to the from complementary technologies.1
political pundit who predicts national election results by At immunology, there has been an explicit editorial pol-
polling a handful of opinions in the pub – tantalizing icy to solicit and encourage papers, especially those that
fun, but not a hard or reliable science. are methodological in nature, that seek to fill some of
Imagine now how well the same pundit might predict these gaps, and we continue to do so. None of us are
trends if they could not only readily poll the full answers born with the ability to eyeball 100 000 lines of text and
and every opinion of several million people, but also inte- spot the answers, which leaves us in need of new tools
grate these answers with several other similarly sized data that can help us extract the meaning. Development of this
sets – their health, employment, spending, hobbies and new toolbox has seen changes in wet-lab technolo-
education records, maybe their family tree too. The catch gies – next-generation genomic sequencing, RNA-sequen-
here is the use of the word ‘integrate’ and what we really cing, mass cytometry, new approaches for TCR repertoire
mean by it: like other branches of science, immunology is amplification – necessarily going hand in hand with the
having to play catch-up with its own technologies to development of new dry-lab software and pathway analy-
understand how, faced with a computer carrying 100 000 sis.
or more lines of data on the question asked, one not only A number of exciting new papers have considered the
imposes some order on that data set in terms of trends tools available to use these omics data sets in various
and pathways, but then works out how to marry this data branches of immunology. One of the pacesetters in how

ª 2018 John Wiley & Sons Ltd, Immunology, 155, 1–2 1

Editorial

to do this for elucidation of phenotype-related biosigna- able to facilitate database searches that can truly equate
tures across omics platforms has long been Damien given TCR sequences and features to the candidate com-
Chaussabel. He points out that there are now more than plexes of peptide–major histocompatibility complex that
90 000 openly accessible data series for analysis within they may recognize – a step that offers paradigm-shifting
the NCBI Gene Expression Omnibus alone.2 This lays rewards for translational medicine.7
down a substantial challenge to the collective brain wat- Meanwhile, there is growing confidence that, whether
tage of present and future researchers if we are to do jus- for vaccinology, infectious disease or tumour immunol-
tice to the required data mining. It is a great opportunity ogy, there is genuine, translational value in bioinformatics
to build careers and also a great equalizer of global approaches to the elucidation of diagnostic or predictive
opportunity: important discoveries are out there, not just transcriptomic biosignatures.8,9 However, as Gliddon and
for the student working in the best-equipped research colleagues have pointed out, point-of-care technologies
institute, but also for the student who can simply go for transcriptomics of clinical samples are still limited,
online and mine the data in a new and creative way. although we await the emerging technology platforms to
Rachael Bashford Rogers recently guest edited a the- turn these approaches into routine diagnostic tests.10
matic review series on new approaches to high-through- It may take many years to fully mine the high-through-
put immune receptor sequencing and its analysis. Within put data that is already out there, without ever funding
immune receptor curation, we now consider not just B- another project. The challenge is to ensure that we train a
cell receptors (BCR) and TCR, but also approaches to sufficient number of high-calibre, bioinformatically liter-
natural killer cell receptor sequencing to understand the ate graduates to rise to the occasion.
functionality of the killer immunoglobulin-like receptors
repertoire.3
Improved technologies for high-throughput BCR
References
sequencing and the ability to link this to functional and
structural studies are having massive impact on infectious 1 Arnold KB, Chung AW. Prospects from systems serology research. Immunology 2018;
153:279–89.
disease and autoimmunity research. Bashford Rogers con- 2 Chaussabel D, Rinchai D. Using ‘collective omics data’ for biomedical research training.
siders the examples of multiple sclerosis, systematic lupus Immunology 2018; 155:18–23.
erythematosus and rheumatoid arthritis, which show 3 Colluci F, Traherne J. Killer-cell immunoglobulin-like receptors on the cusp of modern
immunogenetics. Immunology 2017; 152:556–61.
characteristic BCR repertoire skewing and some evidence 4 Bashford-Rogers RJM, Smith KGC, Thomas DC. Antibody repertoire analysis in poly-
of shared repertoire defects, with clear implications for genic autoimmune diseases. Immunology 2018; 155:3–17.
therapeutic approaches.4 From BCR repertoire analysis in 5 Ghraichy M, Galson JD, Kelly DF, Tr€ uck J. B-cell receptor repertoire sequencing in
patients with primary immunodeficiency: a review. Immunology 2018; 153:145–60.
primary immunodeficiencies, several findings are com- 6 Izraelson M, Nakonechnaya TO, Moltedo B, Egorov ES, Kasatskaya SA, Putintseva EV
mon across diverse primary immunodeficiencies, arguing et al. Comparative analysis of murine T-cell receptor repertoires. Immunology 2018;
that BCR repertoire characteristics can offer clinically 153:133–44.
7 Glanville J, Huang H, Nau A, Hatton O, Wagar LE, Rubelt F et al. Identifying speci-
valuable indicators of the functionality of the B-cell com- ficity groups in the T cell receptor repertoire. Nature 2017; 547:94–8.
partment, whatever the underlying defect.5 8 Chaussabel D, Pulendran B. A vision and a prescription for big data-enabled medicine.
The TCR sequencers have faced a similar but different Nat Immunol 2015; 16:435–9.
9 Li S, Rouphael N, Duraisingham S, Romero-Steiner S, Presnell S, Davis C et al. Molec-
path in terms of development of the wet-lab approaches ular signatures of antibody responses derived from a systems biology study of five
to obtain quantitatively faithful representations of the human vaccines. Nat Immunol 2014; 15:195–204.
rearranged repertoire, and then the computational 10 Gliddon HD, Herberg JA, Levin M, Kaforou M. Genome-wide host RNA signatures of
infectious diseases: discovery and clinical translation. Immunology 2018; 153:171–8.
approaches to curate and interpret this data.6 An addi-
tional dimension here is the need to develop algorithms

2 ª 2018 John Wiley & Sons Ltd, Immunology, 155, 1–2