By Dr. Ruth Bowness
Interview with Ramit Mehr, Professor of the Mina and Everard Goodman Faculty of Life Sciences at the Bar-Ilan University who was named a fellow of the Society for Mathematical Biology in 2019.
You were named the Fellow of the Society for Mathematical Biology in 2019 for your distinguished contribution to the discipline and also contributions to the Society. Could you first tell us about your research background, and how you arrived in your current position?
I did my BSc in Physics and my MSc in condensed matter Physics, at a time when the study of complex systems – actually addressing the complexity rather than simplifying every system to a low-dimensional model – became fashionable. Think Fractals, Chaos, self-organizing systems, emergent patterns and so on. Those were exciting times, and I was drawn to simulating such complex systems and solving the algorithmic challenges this posed. So when looking for a PhD topic, I looked a bit wider – not only in physics but also chemistry and applied math; and the research that finally caught my interest was the work done in mathematical immunology by Zvia Agur and the late Lee Segel. I ended up doing my PhD on “Mathematical Modelling of T Cell Development in the Thymus” under the supervision of Prof. Amiela Globerson from the Department of Cell Biology and Prof. Lee Segel, then in the Department of Applied Mathematics and Computer Science in the Weizmann Institute. By the time I submitted my PhD thesis, I was already a de facto postdoc with Alan Perelson at the Theoretical Biology & Biophysics group in the Los Alamos National Laboratory, NM, USA. From there I went to Princeton University, to work as a research associate in the Molecular Biology department and learn more immunology, prior to obtaining a tenure-track position in Bar-Ilan University, Israel – the first university in Israel to open a computational biology undergrad study program.
What attracted you to computational immunology?
It was the complexity of the immune system that attracted me, combined with the fact that I saw where modeling and simulations can make a contribution. The immune response involves cells of various types, including B, T and NK lymphocytes expressing a large diversity of receptors which recognize foreign antigens and self-molecules. These cellular repertoires interact through a complicated network of communication, regulation and control mechanisms. This is what enables the immune system to perform the functions of danger recognition, decision, action, memory and learning. The dynamics of immune cell repertoires are, as a result, highly complex and non-linear. Over the years, I developed (along with my students) mathematical models and computer simulations, and later some novel bioinformatical methods, in order to analyze the dynamics of the immune system in various situations and predict the results of experimental and medical interventions.
What is your favorite research paper (by another mathematical biologist)?
It’s really hard to choose, as I there were so many I liked over the years! If I have to choose one, though, I’ll use the opportunity to make a tribute to the late Lee Segel. My favorite paper of his, which I sometimes use in teaching, is Segel and Lev Bar-Or’s “On the Role of Feedback in Promoting Conflicting Goals of the Adaptive Immune System” (The Journal of Immunology, 1999, 163: 1342–1349). It is an excellent example of an exploration of ideas using mathematical models, a practice which is becoming rare in the current era of data-driven modeling.
Have you encountered any surprising results in your research?
Many times! I could write pages and pages about the times when “current models” as presented to me by immunologists were unable to explain the available data, and new hypotheses had to be formed and tested, and found to explain the data – hypotheses which were later validated experimentally! To mention only a few, there was the discovery of feedback effects in T cell development, which also had interesting implications to HIV infection dynamics and to immune system aging; the discovery of phenotypic reflux in B cell development, which made it possible for us to quantify the population rate of receptor editing; the explanation of allelic exclusion as a probabilistic phenomenon; the explanation of repertoire shift; the discovery that the enzyme UNG may be over-expressed in ectopic germinal centers in myasthenia gravis patients; and many more.
What is something exciting that you are currently working on?
In a collaboration with Dr. Meirav Kedmi, head of the lymphoma service at the Sheba Medical Center, we study B cell lymphomas, and have recently managed to track the malignant clone in the patient’s peripheral blood. High-throughput sequencing of the clone’s immunoglobulin variable region gene (IgV gene, it’s unique “ID”) alone does not suffice in this case, as such clones keep mutating their IgV genes; we could only do this using a combination of high-throughput sequencing and lineage tree analysis to elucidate the relationship between the original malignant clone in the diseased LN and the further mutated malignant clone(s) in the peripheral blood.
What is your favorite research paper that you have written?
Again, it’s hard to choose! Besides, like all PIs, I’m always more excited about the current work than about what’s done and published. So I leave paper writing to the students and postdocs[Symbol]. Of the papers I’ve written myself, I think the three from my work in Princeton (all of which were published in 1999 in The Journal of Immunology, which was at the time the highest-impact immunology journal that accepted modeling papers) were the most fun to write.
How have you found working with experimentalists?
In most cases, it took a while to develop a common language, but when collaborators were enthusiastic and responsive, it was always a pleasure. I certainly learned a lot from all my collaborators.
What do you foresee as the biggest challenges in modeling of immune system function?
The study of lymphocyte repertoires will eventually enable clinical immunologists to develop better therapeutic monoclonal antibodies, vaccines, transplantation donor-recipient matching protocols, and other immune intervention strategies. The more we learn about lymphocyte repertoires, however, the more we appreciate the staggering complexity that underlies their generation, selection, and function. When I started working in this area in the beginning of 1991, molecular markers and methods for investigating lymphocyte development and behavior were just being developed, and the human genome project was in its infancy – it has just presented as a possible plan to the US congress. During the years of my work in the field of theoretical immunology, I have seen it grow from a small group of interested individuals to a rich, active and challenging research field, whose members are becoming better integrated within the general immunology community. Theoretical immunology is still growing and has not yet fulfilled its potential, however. The recent development of high-throughput methods for repertoire data collection – from multicolor flow cytometry through single-cell imaging to deep sequencing – presents us now, for the first time, with the ability to analyze and compare large samples of lymphocyte repertoires in health, aging and disease. This has a huge potential for identification of subtle defects or changes in immune function, and developing between vaccines, better interventions in autoimmune diseases and malignancies, and better ways to rejuvenate the immune systems of elderly people. The exponential growth of these datasets, however, challenges the theoretical immunology community to develop methods for data organization and analysis. This task is orders of magnitude more difficult than standard sequencing and genomic analysis. First, there is the repertoire complexity itself, which means that one cannot use “reference genes” in the analysis, and the available computational tools are of no use for theoretical immunologists; research groups must struggle to create the correct experimental controls and computational tools, as in the software tools we developed for Ig gene sequence data analysis. The analysis of B cell repertoires is even more complex than that of T cell repertoires – because of the additional diversifying processes that B cells undergo, i.e. somatic hypermutation and isotype switch. Furthermore, the need to test hypotheses regarding immune function, and generate predictions regarding the outcomes of medical interventions, necessitates the development of complex mathematical and computational models, covering processes on multiple scales, from the genetic and molecular to the cellular and system scales. The theoretical immunology community has by now internalized the lesson that the immune system – or even any given part of it – are too complex to enable system behavior prediction out of highly simplified models. Hence the rising popularity of multi-scale modeling in this community. However, we need not only multiscale but also multi-clonal models to truly model immune repertoire behavior.
What impact has your research had outside of academia?
My earlier research has mostly been in basic immunology – addressing lymphocyte development, repertoire development and selection – so that clinical applications were only sidetracks, as in the case of applying the lessons learned about feedback in T cell development to better understand the dynamics of HIV infection. Later on, and especially as the use of lineage trees as models for IgV gene diversification has taken over most of the activity in my lab, many of our research projects were driven by clinical questions and data brought by our various collaborators, and have thus made a contribution to the understanding of immune system aging, its function in several autoimmune diseases, and the behavior of malignant B cell clones. However, only recently has any project had immediate clinical impact. I find this very exciting; however, my original interest in the basic issues of lymphocyte repertoire development and selection has not decreased, and I intend to pursue both directions going forward.
What advice would you give to a junior mathematical biologist?
Find good experimentalist collaborators – good ones will not only supply experimental data to play with, but will also keep your model assumptions grounded in what’s feasible biologically, force you to communicate your model and results clearly, and introduce you to more potential collaborators.
What is the best part of being a computational immunologist?
Being paid to have fun! As far as I am concerned, that’s the best part of being a scientist in every discipline: we’re paid to follow our “intellectual nose”, find and solve interesting problems. Which is fun, or we wouldn’t be doing it, right?
Finally, what do you do in your spare time?
I enjoy spending time with my kids and grandkids, go to see theater shows with my partner, and read a lot. Favorite authors: Terry Pratchett, Ursula LeGuin. Favorite comics: PhD Comics, obviously!