New publication: Mathematical modelling for antibiotic resistance control policy: do we know enough?

Newly published – a really nice review on AMR modelling – I have had really very minimal input – Gwen Knight and other authors have done all the heavy lifting on this!

Knight GM, Davies NG, Colijn C, Coll F, Donker T, Gifford DR, Glover RE, Jit M, Klemm E, Lehtinen S, Lindsay JA, Lipsitch M, Llewelyn MJ, Mateus ALP, Robotham JV, Sharland M, Stekel DJ, Yakob L and Atkins KE 2019. Mathematical modelling for antibiotic resistance control policy: do we know enough?. BMC Infect Dis 19: 1011.



Antibiotics remain the cornerstone of modern medicine. Yet there exists an inherent dilemma in their use: we are able to prevent harm by administering antibiotic treatment as necessary to both humans and animals, but we must be mindful of limiting the spread of resistance and safeguarding the efficacy of antibiotics for current and future generations. Policies that strike the right balance must be informed by a transparent rationale that relies on a robust evidence base.

Main text

One way to generate the evidence base needed to inform policies for managing antibiotic resistance is by using mathematical models. These models can distil the key drivers of the dynamics of resistance transmission from complex infection and evolutionary processes, as well as predict likely responses to policy change in silico. Here, we ask whether we know enough about antibiotic resistance for mathematical modelling to robustly and effectively inform policy. We consider in turn the challenges associated with capturing antibiotic resistance evolution using mathematical models, and with translating mathematical modelling evidence into policy.


We suggest that in spite of promising advances, we lack a complete understanding of key principles. From this we advocate for priority areas of future empirical and theoretical research.

International grant panel experience

Last week I had the wonderful experience of participating in an international grant panel – this time for NSERC in Canada. Being on an international panel showed me new aspects of grant writing, reviewing and decision making. Specifically:

1. Being on an international panel means that we experience people with different grant panel cultures. These throw up quite varied views: how much do you value the publication track record of the applicants vs the actual proposed research in the grant? How crucial is preliminary data? How important is the training and experience of people working on the grant (the Canadian system uses the term ‘HQP’: highly qualified people)? These seem to have different emphases for people used to the UK system, or EU, or USA, or Canadian.

2. How to disagree with dignity. One of our European panelists was exemplary in the way in which he surfaced differing views, while remaining calm and rational at all times, asking tough questions of other panelists where he took a different view, but also backing down graciously when it was clear that was needed for consensus building.

3. The importance of response to reviewers. I know this is important with UKRI grants – but it really came across profoundly in this process for two reasons. (i) As a demonstration of the depth of thought and knowledge that sits underneath the material in the proposal itself, which by necessity is always edited down. (ii) As a test of the strength of the team. This is especially evident in interdisciplinary research, where the PI relies on expert answers from the whole team: can the PI pull together answers to all questions on a short time frame? Sometimes that requires high levels of commitment from team members – which may or may not be manifest.


Suitable offices for computational biology

Last week my research group moved to a new office – a lovely office for 6 – adjacent to the tea room – and with a move of building that will bring us closer to the animal scientists (cows and chickens) and agricultural economics modellers (whom I have found out use many of the same Bayesian techniques that we do) – though sadly moving us out of the building we share with our environmental science, veterinary science, and bioinformatics colleagues (it is only next door).

First, thanks go to our Head of School for listening to our concerns and facilitating the move. This move is very important for my group, and there is history to it, and important things to learn.

For the past 8 years, my group has been located in a very large open plan office, shared with very many other students and postdocs, the overwhelming majority of whom are laboratory based, and so use their desks on an occasional basis. This means that there are a lot of people, movement, distractions and noise. Even when the building was being planned, the computational biologists at the time (not my group – members of the Centre for Plant Integrative Biology – who were originally going to be in the space) raised their concerns that this would  be a space in which they would be unable to work effectively. While their views were represented by a number of people – including myself – the University’s former Head of Estates was not prepared to listen to them or us – and insisted in creating the space as a very large open plan area.

My group moved in about 8 years ago, and since then, the majority of my PostDocs and PhD students have found it very difficult to work there – and so have worked away from the office. The situation at the moment is typical: of my current group, 3 out of 4 find it very difficult to work in the space, and so mainly work from home. The consequence is that my group rarely communicate (outside of formal meetings), do not benefit from informal peer support, from each other, or from other students/researchers, and this has had a noticeable impact on their well-being. Here is the key point:

Computational biologists have different needs from their offices from laboratory based researchers.

(i) Computational biologists are 100% desk based so have a different relationship with that working space from laboratory people who only use their desks occasionally

(ii) Computational biology desk work is mentally intense – writing computer code and algorithms – so needing intense concentration. A location with a lot of movement and noise (people coming in and out of their lab) is not one where we can work effectively.

(iii) Many computational biologists (myself included) are Highly Sensitive and Introverted (capital letters intended). This gives us our strengths – e.g. the abilities to write complex algorithms, or see patterns in large volumes of data, or work effectively across two or more disciplines – but also means that we are much more sensitive to movement, noise, distraction and large numbers of people than more ‘neurotypical’ people (whatever than means). Many computational biologists need quieter, calmer places to work (and rest), especially when we need to concentrate on what we do best.

I am extremely happy that our Head of School got this – but am somewhat surprised by the number of people who have responded negatively to the move and reasons for it. Another way to think about it is that laboratory scientists have very specific needs for the different types of laboratories they need (e.g. a microbiology lab has to have certain containment requirements to protect people from pathogens; a mass spectrometry lab needs a certain level of cleanliness to protect the samples etc), and research organizations are always willing to build research laboratory spaces to meet these needs. Well, computational biology labs also have requirements: for quiet concentration and connection, best met by smaller shared offices (we do need other people to talk to – just not too many at once!), with access to a common space for longer conversations – and these needs should be taken just as seriously.

When I went into our new group office last week, I saw two of my PhD students sitting at their new desks, screens full of code, and have never seen them look so happy!

On the train to ARAE2019 – climate change friendly conference attendance

I am on my way to Tours by train for the 8th Symposium on Antimicrobial Resistance in Animals and the Environment in Tours, France. I am excited by this for a number of reasons.

1. I have resolved not to fly in the UK or in Europe but to travel by train or other means. We academics have very bad habits with regards international travel – frequently flying to conferences or collaborators. I try to keep my international travel to a minimum, and even then have had recent long-haul trips to China, South Africa and Canada. While it would be difficult cut out international travel altogether – it is an important part of our job – I think we do need to be more mindful of the impact of our behaviour on climate change. It is perfectly possible to travel within Europe by train – and not especially inconvenient. (This journey will involve four trains: Leicester->London, London->Paris, Gare du Nord->Gare Montparnasse; Paris->Tours) but not much more inconvenient than getting to/from airports and the trains are really very pleasant.

2. I have also made a conscious choice to attend a European rather than further afield conference. I had originally planned to attend EDAR5 in Hong Kong – having greatly enjoyed EDAR3 and EDAR4 – and knowing full well that Tong Zhang and his committee will have put on an amazing event. But ARAE8 has some advantages too – much nearer (no long haul flight = less carbon) and it is a group of people I mainly do not know – so I am really looking forward to meeting some new people and to hear new and different perspectives – even if it means that because they do not know me, I will be presenting a poster rather than talk.

3. First opportunity for my PhD student Henry Todman to present at an international conference. He too will be bringing a poster (for which he won second prize within the Division of Agriculture and Environmental Science at our Campus wide postgraduate conference last week).


Grant Committee Reflections – what I learned

Following on from my previous post, last week I had my first experience of being on a BBSRC committee, and it was certainly interesting. Main points:

1. The committee takes its responsibility seriously. IMs take time to read and understand the grants – even when on topics outside of expertise – and try to form fair and objective judgments.  The process – while time-consuming (and expensive) is also fair and robust. The discussions were respectful – even when there was disagreement, the panel tried to form objective judgments based on the grants in front of them.

2. There is a scoring culture that new committee members have to adapt to. I was unsure how to calibrate from the 6 point scale the reviewers get to the 7 point scale the committee has – what was interesting is that there is a clear shared culture over what the scores – and especially in range from 5 to 6 – ‘mean’. So this means that IMs do tend to decide their own scores for grants – using the referee scores as a rough guide – rather than taking an average of referee scores. This is actually a good thing for two reasons. First, it brings consistency – if the committee members work to a mutual understanding of the scores, then they can score consistently, in a way that referees cannot, because their scores are much more individual and biased. Second, it means that what the referees actually write is more important than the score they give. A high score without clear justification is not worth much to the grant; a lower score, but with questions that are well answered by the grant authors, will not be adverse to the grant – indeed well-answered questions will help it.

So this means for reviewers: write good reviews. Whether your scores are high or low, make sure you have clear reasoning behind your score, and clear questions for the grant authors. For grant writers: don’t get too fixated on the actual reviewer score. Focus on the comments/questions and provide well-evidenced answers.

3. Most grants will have at least one IM who is not directly expert in its content. This means that it is really important for grant writers to have sections of the grant (why is the work important and what will it lead to) that are understandable by a broader range of scientists. The work plan should contain technical details – but if the whole grant is technical, it can be hard for IMs to ‘sell’ the grant to each other or the committee chairs.

4. There is a very embedded culture around preliminary data – that was clearly more manifest in the empirical biology members of the panel over the computational biology members. On some levels this is important, but on other levels perhaps it is given too much weight – to the detriment of more innovative research. Of course, it is crucial to provide evidence to support the hypotheses/assumptions of the work plan and choice of research activity. Equally, it is crucial to demonstrate that the lab can be successful in using the particular techniques, organisms/cells etc required for grant success. Without these a grant will not be successful. On the other hand, I do wonder how much value it truly is: preliminary data will have been produced by PhD students or PostDocs who most likely will no longer be in the lab by the time the PostDoc on the project starts. In the end it is the quality of that PostDoc that will make the biggest difference – their background knowledge, their ability to learn new skills, etc. A good postdoc will learn new techniques and skills; a poor postdoc will not – and none of that relates to the quality of the preliminary data. That is not factored into the entire grant process!

5. The funded grants were really awesome. It is a competitive process – and while there is some element of luck – the ones that reached the top were exceptionally good, and  many that will not be funded were also excellent.


Research Associate/Fellow in advanced statistical modelling and machine learning applied to epidemiology of bacterial infection and antimicrobial resistance (fixed term)

Applications are invited from research scientists for the above post in the School of Veterinary Medicine and Science to join an exciting project dealing with antimicrobial/antibacterial resistance (ABR).

The successful candidate will work closely with an interdisciplinary and international team of academics and industrial partners. The project offers a unique combination of expertise in machine learning, statistical and mathematical modelling, bioinformatics, sequencing, cloud computing, microbiology, infection control, food safety, surveillance, epidemiology.

The applicant must have a PhD (or be very close to completion), preferably in epidemiology or data science.  Alternatively, the candidate must have a PhD (or be very close to completion) in statistics, mathematics, machine learning, or computational biology. In-depth expertise in the use of advanced statistical modelling and machine learning for data analysis in biological problems, preferably related to epidemiology of infectious diseases and antibacterial resistance is essential. Strong proficiency in programming/software development is required. Experience/expertise in the following subjects would be desirable: use of deep learning for data analysis, use of advanced statistical modelling and machine learning for the analysis of bacterial whole genome sequencing data and metagenomic data; development of diagnostics or forecasting tools; development of surveillance/monitoring solutions for infectious diseases.

The School of Veterinary Medicine and Science is committed to diversity and equality of opportunity.  The School holds a Bronze Athena SWAN award in recognition of its commitment to equality and diversity and advance the representation of women in veterinary medicine.

This is a full time, fixed-term available immediately until 31 December 2021. Job share arrangements may be considered.

Informal enquiries may be addressed to Dr Tania Dottorini, email: note that applications sent directly to this Email address will not be accepted

Our University has always been a supportive, inclusive, caring and positive community. We warmly welcome those of different cultures, ethnicities and beliefs – indeed this very diversity is vital to our success, it is fundamental to our values and enriches life on campus. We welcome applications from UK, Europe and from across the globe. For more information on the support we offer our international colleagues, visit;

Closing Date
Monday, 1st July 2019
Job Type
Veterinary Medicine & Science
£27025 to £32236 per annum, (pro rata if applicable) depending on skills and experience (minimum £30395 with relevant PhD).

A trend towards more authors on my publications

I’ve noticed that I seem to have more papers with more authors – so I thought I would check to see if that trend is borne out in the data. Here is a graph of the number of authors in all my publications, against year of publication:


With a quick and dirty linear regression trendline: gradient is 0.169 per year with a p-value of 0.00733. So there is statistical evidence to support an increase in the mean number of authors per paper over the 24 years I have been publishing, from a mean of about 2 authors per paper in 1995 to a mean of about 6 authors per paper now  (if you pretend that a linear regression is correct for these data – which it isn’t). Notwithstanding doing a better statistical job, I am happy to accept that the linear regression result is fair: I am writing papers with more co-authors. But why?

First, perhaps, is increasing depth of interdisciplinarity. All my papers are interdisciplinary – and many have authors from different disciplines – even my very first paper was written with academic medical doctors. But as my career has progressed there is increasing complexity – papers that describe research that includes both experimental and modelling work, or papers which contain three or more disciplines (not just two). Related to that is that I also tend to consult more widely about my work – so tend to have more authors who have contributed ideas, discussion, or secondary or tertiary supervision of students.

Second, perhaps, is that I am more open in my view of who becomes an author. Inclusion as authorship those colleagues who have contributed ideas or discussion or small elements (e.g. a database submission) can be a marginal decision (there has to be a boundary somewhere). In the past, I have tended to exclude, now I tend to include.

And so what? Actually, I feel happy and positive when I view this trend. I do not see myself developing as a ‘lone scholar’, but rather as someone who values working across boundaries and disciplines, including a wider range of people into my research. The problems we are tackling – especially antimicrobial resistance – needs this broader approach to be successful.