Gianola et al. Genetics Selection Evolution (2022) 54:9 https://doi.org/10.1186/s12711-022-00704-y Ge n e t i c s Se l e c t i o n Ev o l u t i o n EDITORIAL Open Access Rohan Fernando: a road from Sri Lanka to Ames Daniel Gianola1*, Rodolfo J. Cantet2, Jack C. M. Dekkers3 and Miguel Pérez‑Enciso4 About one year ago, some of us learned that Rohan Fer- Fresno. He carried out postgraduate study at the Univer- nando had announced his intention to retire from his sity of Illinois at Urbana-Champaign, where he obtained position as Professor of Animal Science at Iowa State Masters and Ph.D. degrees, with thesis work focusing University. We felt that it would be appropriate to prepare on methodology in populations undergoing assortative a volume honoring his important contributions to quan- mating and selection. He then continued his career as a titative genetics, especially to animal breeding. With that tenured faculty member, first at the University of Illinois idea in mind, many of his collaborators and colleagues at Urbana-Champaign (1985–1996) and as Professor at (past and present) were contacted and the idea was Iowa State University thereafter. received with great enthusiasm. Genetics Selection Evolu- He has been a productive scholar, with close to 15,000 tion (GSE) was considered to be very fitting as outlet and citations and an h-index of 57 in Google Scholar; excel- Jack Dekkers helped greatly to facilitate the task of pub- lent marks in a small field. His six most cited papers lishing this collection of papers, prepared by current and include his seminal work on best linear unbiased pre- past associates and students. The series touches many diction for marker assisted selection, his pioneering of of the areas where Rohan’s work was influential, such Bayesian methods in animal breeding (at a time when as estimation and prediction in populations undergoing Bayesianism was considered heretical), clarification selection, Bayesian methodology, theory and methods for of the role of markers in Bayesian regression models, a finite number of loci situations, genetic evaluation in including highlighting the importance of genetic similar- crossbreeding populations, prediction with high-dimen- ity in prediction, and development of new methods for sional genomic data, statistical computing of large data Bayesian variable selection. He also was a coauthor of the sets, and software development. Each of the papers was first paper pointing out the potential contribution of ker- subjected to the type of refereeing that is standard for nel methods to genome-enabled prediction. GSE. Rohan is a coauthor of some of the papers, without In addition to many contributions in the areas men- realizing that they, ultimately, would appear in a collec- tioned above, Rohan’s deep insights into difficult prob- tion recognizing his work! lems, coupled with his ability to arrive at elegant and Professor Fernando was born in Sri Lanka in a family convincing solutions, made him an ideal person to col- with strong connections with agriculture, so he devel- laborate with, and to engage in discussion. During his oped an interest in livestock. He attended Aquinas School career, he has assisted hundreds of students and collab- in Colombo, a Catholic junior college, and then came to orators, in a passionate and firm, albeit generous, quiet the United States, where he completed a Bachelor of Sci- and humble, style. Those of us who have had the privi- ence in Animal Science at California State University in lege of counting Rohan as a friend know that behind his sparse rhetoric, there is a person with very high ethical and spiritual dimensions. He has been extremely help- *Correspondence: gianola@ansci.wisc.edu 1 ful to many, representing a sort of "Sister Theresa of University of Wisconsin-Madison, Madison, WI, USA Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Gianola et al. Genetics Selection Evolution (2022) 54:9 Page 2 of 6 Calcutta-equivalent" of animal breeding; always giving, models into thinking in terms of joint, conditional, and seldom requesting. marginal distributions. When Sotan Im and Jean-Louis Apart from being an avid reader of biblical material, he Foulley (INRA, France) came to Illinois for a sabbatical, has always enjoyed company with his friends and fam- both being powerful mathematicians, our mental entropy ily, good food and wine, and traveling, and he has mas- about statistical theory and its relationship to quantita- tered several “foreign languages”, such as C +  + , Matvec, tive genetics gradually began to dissipate. I believe their and more recently Julia. With a nice smile, he could tell a contribution was crucial in Rohan’s development as a friend "Go to hell" (he never did it, though), and make the quantitative scientist, and we finally understood the limi- recipient of the statement quite happy. We wish him luck tations and shortcomings of Henderson’s paper on BLUP in his next assignment where he will enjoy life and love in under selection. It took us a few years of discussion, even a land of oranges, wine, and electric cars, California, but with Henderson, who was a recurrent visiting professor beware of the earthquakes! at the University of Illinois, to get to that point. Since he has been, in some sense, a "best seller" in our When I returned from a sabbatical in Jouy-en-Josas in field, below are excerpts from statements made by per- 1982, I was fully convinced that the Bayesian approach sons whose life was influenced in one way or another by would provide a solution to almost any problem in Rohan. The order of their comments is approximately quantitative genetics. In fact, it has been used even to chronological, trying to match various stages of his devel- evaluate the statistical plausibility of the existence of a opment as a scientist. After reading these remarks, we deity (not surprisingly, the prior is heavily influential). encourage you to study the papers that are included in I narrated my Bayesian epiphany to Rohan, an experi- this series through which Rohan’s influence on the ideas ence that was equivalent to defending an innocent per- presented in the collection will become evident. son in a trial organized by the Spanish Inquisition or the Ku-Klux-Klan. We found common ground and a paper Daniel Gianola (University of Wisconsin, United emerged on “Bayesian stuff for animal breeders”. Without States) my interactions with Rohan, and with Daniel Sorensen, I My first two guinea pigs (also known as graduate stu- would have never understood Bayes in a proper manner. dents) at the University of Illinois at Urbana-Champaign I have known Rohan for more than four decades and he (1978) were Wilson Nelson Mulenga Mwenya and Rohan has been an extremely important influence in my life as a Luigi Fernando. Both were excellent but different from scholar. I could narrate many anecdotes, but space is lim- each other in interests. Mwenya became a Dean at the ited. However, I would like to end the statement with the University of Zambia, whereas Rohan chose scholarship. following sentence: "If you were a coach, he should be in At that time, I had developed a "monster course" that your starting five. If you do not love the man, go and see a covered essentially everything that was not being taught shrink: you may be ill". elsewhere, and with an important coverage of BLUP (best linear unbiased prediction) a very hot, mandatory, and mystical topic at that time. Most of us could follow Hen- Katherine Hanford and Stephen Kachman derson’s algebra, but without understanding the implica- (University of Nebraska‑Lincoln, United States) tions. Mwenya and Rohan took the course and both did Rohan has influenced our lives in many ways, both per- well. A problem, however, was that Rohan would not ask sonally and professionally. We have been friends and questions in class (typical of him) but would later show colleagues of Rohan since we met as graduate students up in my office with questions for which I did not have in Daniel Gianola’s lab at the University of Illinois at a readily available answer. To get rid of this "nuisance Urbana-Champaign in 1981 and, over the past four dec- parameter", I would give him arcane books on math- ades, Steve and Rohan have continued their collabora- ematics and linear algebra and command: "read this and tion. We have numerous photos of Steve and Rohan you will understand". That did not work either, because sharing a laptop or writing on a white board or a sheet of he would return with questions about the books, so my paper, working on problems ranging from how to imple- problems got worse instead of better! Eventually, we ment a statistical method in software, deciphering a bit helped each other understand important foundational of statistical theory, incorporating genomic information issues from our field, some of which we continue to dis- into genetic evaluations, and the pros and cons of Rohan’s cuss. Apart of the mentor–mentee relationship, we later latest favorite programming language. Whether spending became colleagues and friends. time at Iowa State University, serving together on various At that time, graduate students in animal breed- committees and projects, or getting together at meetings, ing were beginning to take classes in mathematical sta- we always looked forward to the opportunity to spend tistics, and we gradually evolved from writing linear time with Rohan, where Steve could be assured to leave Gianola et al. Genetics Selection Evolution (2022) 54:9 Page 3 of 6 with a new idea to pursue and Kathy would be caught up of a thesis (Leah Dodell, "Revisiting Biblical Games in a on Rohan’s kids. Bayesian Framework", Ph.D., 2013, Emory University) in my congratulations on Rohan’s retirement. Rodolfo J. C. Cantet (Universidad de Buenos "In this paper, I revisit a few of the most debated tales Aires‑INPA‑CONICET, Argentina) from the Old Testament and model them in the frame- Dr. Fernando and I met at the University of Illinois work of Bayesian games. I model three situations - Jacob’s at Urbana-Champaign back in 1986, and he acted as deception of Isaac, G-d’s ten plagues, and Abraham’s sac- my thesis codirector until I finished my PhD in 1990. I rifice - as dynamic games with private information. By have many fond memories of those days. With Rohan, I solving for the Perfect Bayesian Equilibria that occur in learned to address hard research problems in quantita- the Torah, I find conditions that must hold for characters tive genetics and genetic evaluation without much fear. to be willing to take the actions that they do. I also exam- Rohan, with his positive and generous contribution, ine how characters’ actions would have changed if they helped me to learn how to find a solution. Actually, I had held different values. My results shed light on which believe he induced in his students the perception that we interpretations of biblical stories hold the most weight were the ones who had solved the problem, when in fact when characters maintain consistent beliefs and act upon it was him who had led us to that point. them in a sequentially rational manner." A paragraph in Wikipedia’s description of the contri- From someone who retired three years ago, I can say butions of Dr. Fernando refers to the covariance between that I highly recommend it. Best of luck in whatever you relatives for marked quantitative trait loci (QTL) and for plan to do, but do not forget to study the Bible. multi-breed populations. I remember telling Rohan in 1988 about composite breeding and the need to develop Michael Lynch (Arizona State University, United a theory for the covariance between relatives in admixed States) breeds. A conversation of half an hour was enough to set One of the greatest influences of my entire career came Rohan in motion. He solved the problem for additive and in the early 1980s at the University of Illinois at Urbana- additive plus dominance models and proposed an imple- Champaign, when I struck up a relationship with what mentation of the additive model for evaluation of a two- was then one of the strongest groups of theoreticians breed composite. in the world in quantitative genetics: Rohan Fernando, I was impressed by his 1989 BLUP and marked QTL Jean-Louis Foulley, Daniel Gianola, Michael Grossman, paper with Mike Grossman. In that publication Rohan and Charles Henderson. One of the most memorable derived the covariance between additive effects condi- sets of events involved a reading group focused on clas- tional on identity-by-descent (IBD) using his own and sical papers in the field, including Fisher 1918 and going thorough reasoning. He showed how probabilities of IBD even further back to Pearson’s papers. Rohan and Dan set are functions of the recombination rate between a causal a frighteningly high bar by writing highly lucid 20-page variant and the marker gene. This derivation is remark- sets of notes on each paper, which I still have today. Their able and emphasizes an individual mode of transmission presentations provided an extraordinarily clear and rig- of the genome that is affected by the pedigree. The work orous overview of some of the most difficult foundational of Professor Fernando on IBD and covariance matrices of papers in the field. I was able to bring a more evolution- genetic effects has been extensive. ary biology perspective, and jointly we put together an These statements reflect my view of what Rohan’s con- extraordinary couple of semesters with long sessions, tributions were and of what Rohan did for my thinking that probably should have been recorded. and education as a researcher in quantitative genetics. Animal breeders should all be grateful to him for his Miguel Pérez‑Enciso (ICREA, Universidad insight and intellectual generosity. Autónoma de Barcelona, Spain) Professor Fernando always smiles. Rohan is a rather hum- Joel I. Weller (Agricultural Research Organization, ble person who shies away from the center of the stage The Volcani Institute, Israel) and who does not discuss openly in public. Yet, he can I met Rohan when I did my first sabbatical at the Univer- spend hours calmly discussing with a colleague in private sity of Illinois at Urbana-Champaign in 1987–1988. We on a recondite issue he had encountered on a well-known are nearly the same age but could not have more differ- and accepted procedure. His understanding of statistical ent early biographies, even though we both have strong issues in quantitative genetics is simply impressive. religious beliefs. Daniel Gianola called Rohan a “Biblical Strangely for a highly self-contained scientist, up to this Bayesian”, and that is probably a pretty-good description. point, he has not authored a paper without coauthors. I have therefore decided to include the following abstract This may be a coincidence but it may also be a result of Gianola et al. Genetics Selection Evolution (2022) 54:9 Page 4 of 6 Rohan always looking for a partner to share knowledge a good influence on Rohan; I know he has been a good with and to help. influence on me. His prose is terse and sober, as is his personality. I remember the first paper that we wrote together at the Max F. Rothschild (Iowa State University, United University of Illinois at Urbana-Champaign. I presented a States) first draft to Rohan, and we then spent an entire morning It has been a great personal and professional pleasure reviewing it word by word, removing as many as possi- to know and work with Dr. Rohan Fernando in the Ani- ble, and positioning each colon and semi colon as care- mal Breeding and Genetics group at Iowa State Univer- fully as a surgeon would do with a scalpel in a delicate sity these many years. I first met Rohan when he was a surgery. When I would talk to him while he was at Iowa graduate student with Dan Gianola at the University of State University, Professor Rohan was always excited by Illinois at Urbana-Champaign. I always enjoyed his posi- a new programming language, be it C +  + or Julia, talk- tive attitude and of course his interesting insights in solv- ing excitedly about how one could use it to program a ing statistical and genetics problems. While I was leader famous differential equation as elegantly as possible. of the Animal Breeding and Genetics group at Iowa Professor Fernando is author of over 200 publications State University, I was instrumental in getting him and with colleagues from all over the world, including the Jack Dekkers hired and these colleagues have worked authors of this paper. They have covered, over the years, well together and been great collaborators for the group, numerous topics that vary from Bayesian theory to plant department, and around the world. breeding. They all share a clear objective of solving meth- Rohan, as everyone knows, is an outstanding quantita- odological problems, usually employing novel, surpris- tive geneticist and also he has helped to not only develop ing, and highly original solutions. Many of his works have theory but also program solutions for practical problems. had a profound influence on the field. Perhaps his most His contributions in these areas have been significant. influential and premonitory work was "Marker assisted People who also know Rohan know that, while selection using best linear unbiased prediction", coau- extremely helpful and friendly, he does not seek out peo- thored with Michael Grossman in 1989. In that work, he ple, rather they come to him. In this regard, I have often derives the relationship matrix and its inverse when using joked that he sits in his office or “cave” and we need to a marker and a QTL. The key contribution was that he go push him out or join him. I and a few of my students treated the QTL allele effects as random, anticipating have collaborated with him on several occasions and his genomic selection principles by a decade. At a time when efforts greatly improved the research and work of my stu- everybody was looking for candidate genes and excited dents. His is a patient mentor and works hard to explain about marker assisted selection, this work was amply even the most difficult concepts and has successfully lec- cited but perhaps its philosophical implications were not tured around the world. fully understood at that time. Professor Rohan Fernando’s contributions to Animal Breeding and Genetics and statistical applications in our field have been considerable. It is my hope that he will Robert C. Elston (Case Western University, United continue to contribute not only his ideas to the field but States) continue his friendship with his many colleagues here in Rohan spent a postdoctoral year with me in New Orleans Iowa and around the world. in 1991–1992. Chris Stricker was also with me at the same time and I believe this was when they first met; in Dorian Garrick (Massey University, New Zealand) any case, they soon realized they had a common interest Rohan Fernando has always been quiet and unassum- and they became coauthors of at least seven papers. They ing, so his presence could easily go unnoticed. I first met had each come to me knowing that I was developing soft- Rohan Fernando in the mid 1980’s and quickly learnt ware to calculate likelihoods on large human pedigrees that his publications and his presentations were care- and they worked with me to develop new algorithms in fully thought out, from first principles, and always very this area. However, their real interest was in livestock thorough. I immediately had a lot of respect for his work, pedigrees, which have many more inbreeding loops than particularly his 1989 paper on “Marker-assisted selection are typically found in human pedigrees; and so they using BLUP” when it first came out. developed an algorithm that cut the numerous loops, but I did not really get to know Rohan well until I joined which gave a good approximation to the likelihood. him at Iowa State University in 2007 and began work- As with all good students – and Rohan was no excep- ing closely with him. His first big contribution to my tion – there comes a time when it is no longer clear who academic career was in the understanding of Bayes- is the mentor and who the mentee. I hope I have been ian approaches to some animal breeding and related Gianola et al. Genetics Selection Evolution (2022) 54:9 Page 5 of 6 problems, particularly BayesA and BayesB. Most of my of the corridor between his old office and mine. A typi- knowledge on linear model theory and applications was cal scenario would involve me thinking up a hare-brained based on my pre-doctoral interactions with Dr. Robert idea for improvement of an approach and suggesting it Anderson and with Arthur Gilmour during his Ph.D. to Rohan. Sometimes he would immediately and politely studies, and then during my Ph.D. at Cornell from Drs. point out its inadequacies. On a few occasions, when it Searle, Quaas, Van Vleck, Henderson, Pollak, and Cole- warranted further consideration, he would go back to man. I had been taught a fair bit about Bayesian inference his pen and paper, or his whiteboard, and work the issue by George Casella when I was a Ph.D. student at Cornell through. Then, often in the middle of the night, Rohan in the mid 1980’s. But, at that time we did not know how would start playing with small examples. That became a to apply it to real world problems associated with estima- lot easier after 2012 when we started using Julia rather tion of variance components or genetic evaluation. Or at than R to prototype new ideas before C +  + implementa- least I did not know how to do it. tion. Rohan was one of the first to use Julia and has been Daniel Gianola and his students, particularly Alicia responsible for much of its adoption by the animal breed- Carriquiry, made many nice presentations on Bayesian ing community. Communication became even easier for inference at conferences, promoting the appealing con- Rohan when Jupyter notebooks became available. Instead cepts, but they always stopped short of educating me of emailing a snippet of R or Julia code, the norm became as to how it could be done in practice. So, I never came a Python notebook that contained the Latex markdown, away from those activities with any better idea of how the simulation, results, and usually graphs, making it to apply it in real life. My first landmark in understand- much easier to share ideas with other interested parties. ing real-world applications of Bayesian inference was at Reminiscent of my memories of Dick Quaas, most a course that Martin Tanner held in Armidale. There, I of the activities Rohan has worked on have led to use- learned about Markov Chain Monte Carlo, Metropolis– ful extensions of known concepts and much improved Hastings, Gibbs sampling, data augmentation, and other understanding. Those that appear in refereed journals treasures, but mostly applied to problems quite unlike represent only the tip of the iceberg of such endeavours. those involving large-sparse sets of mixed model equa- Rohan almost always reverts to the most basic principles tions like the ones we routinely used in REML estimation when considering a problem. He never takes some of the of variance components, or in genetic evaluation. results others claim for granted. When Rohan referees My next landmark was in reading the 2001 Meuwissen, a novel paper, it means rederiving all the findings in the Hayes, and Goddard paper. I did not spend as much time paper. coming to grips with their BayesA and BayesB methods Rohan has made a remarkable contribution to the lives as I should have, but their so-called BLUP approach was of many students, postdoctoral fellows, and young and as familiar to me as bread and butter. BayesA and BayesB old scientists. I consider myself very fortunate to have would have had a much bigger impact on me at the time crossed paths with him, and to have been able to work if we had real-life SNP chip data, but at that time I was closely with him on a number of problems over the last still working with microsatellite markers. 20 years. Soon after my arrival at Iowa State University, I received my first 50 k bovine genotype data on some Jack C. M. Dekkers (Iowa State University, United 1000 animals and I started working with Rohan on States) how we might make inference. Rohan convinced me of One of the most brilliant yet modest scientists with the merit of BayesA and BayesB and helped me under- whom I have ever had the chance to work with, along stand some of the intricacies of those algorithms. The with one of the kindest and most patient persons that I light-bulb finally turned on when Rohan again patiently have ever met. That is how I would describe Rohan. demonstrated to me the issues of Gibbs sampling and I got to know Rohan after I started as a faculty mem- I realized the very close similarity between single-site ber at Iowa State University in 1997. I had read some of Gibbs sampling and Gauss–Seidel iteration, then the his papers but, frankly, had trouble understanding them close similarity between joint Gibbs sampling and block in full, mostly because I tend to approach problems from Gauss–Seidel. The Gibbs sampler is a most amazing algo- a more conceptual manner, in contrast to the math- rithm. Once I finally understood more about BayesA and ematical and statistical rigor that Rohan employs. Yet, BayesB, we could start thinking about what we might do as we started to discuss problems in animal breeding, to further improve the algorithms and to expand the so- we were able to find much common ground, where our called Bayesian alphabet. approaches complemented each other. Over two decades Rohan moved his office next to mine when Dr. Richard of fruitful and enjoyable collaborations followed, result- Willham retired, so we would not have to pace the length ing in 56 coauthored papers, and still going strong. Gianola et al. Genetics Selection Evolution (2022) 54:9 Page 6 of 6 I have always admired Rohan’s rigorous and meticulous attention to detail when it comes to statistical and quan- titative genetics, always striving to getting to the bottom Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ of the problem and not taking anything for granted. I lished maps and institutional affiliations. remember us visiting for hours in the late 1990’s with Ali- cia Carriquiry’s husband, Wolfgang Kliemann, who was a mathematics professor at Iowa State University. Wolf- gang explained the mathematical basis of the Gibbs sam- pler and the Metropolis–Hastings sampler to us. And not until Rohan had fully convinced himself of their validity, was he willing to let Markov chain Monte Carlo replace, or rather, complement the likelihood-based philosophy that was ingrained into his approach to statistical genet- ics. Since that time, he has never looked back and made, and continues to make, tremendous and novel contribu- tions to the field of Bayesian statistics applied to animal breeding and quantitative genetics. Even now, upon his retirement, he is embarking on the power of machine learning, getting to the bottom of it, convincing himself that it works, connecting it to his knowledge of likelihood and Bayesian methods, and employing it to solve prob- lems in our field. In addition to a being a superb scientist, Rohan has also been a great educator and mentor. The students, post- docs, and visiting scientists who have had the chance to work with him can attest to the close working relation- ship that he developed with each of them. Often, prior to Covid19, when you walked by his office, you would see a student or post-doc sitting next to him behind the computer, as they were working together on a program or a paper. Even now, during the pandemic, lengthy Zoom calls with students and post-docs have been the rule of the day for Rohan. The patience and collegiality that he has exhibited over the years with all graduate stu- dents and post-docs who he worked with, regardless of whether they were his own or not, is truly remarkable and enviable. Rohan, my friend and colleague, enjoy your well- deserved retirement but, please, stay in touch. Authors’ contributions DG, JD, RJC, and MPE contributed equally to the paper. All authors read and approved the final manuscript. Ready to submit your research ? Choose BMC and benefit from: Competing interests • fast, convenient online submission The authors declare that they have no competing interests. • thorough peer review by experienced researchers in your field Author details • rapid publication on acceptance 1 University of Wisconsin-Madison, Madison, WI, USA. 2 Universidad de Buenos • support for research data, including large and complex data types Aires-INPA-CONICET, Buenos Aires, Argentina. 3 Iowa State University, Ames, IA, USA. 4 ICREA, Universidad Autónoma de Barcelona, Barcelona, Spain. • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions