Services-Based Systems Architecture for Modeling the Whole Cell: A Distributed Collaborative Engineering Systems Approach

Abstract

Modeling the whole cell is a goal of modern systems biology. Current approaches are neither scalable nor flexible to model complex cellular functions. They do not support collaborative development, are monolithic and, take a primarily manual approach of combining each biological pathway model’s software source code to build one large monolithic model that executes on a single computer. What is needed is a distributed collaborative engineering systems approach that offers massive scalability and flexibility, treating each part as a services-based component, potentially delivered by multiple suppliers, that can be dynamically integrated in real-time. A requirements specification for such a services-based architecture is presented. This specification is used to develop CytoSolve, a working prototype that implements the services-based architecture enabling dynamic and collaborative integration of an ensemble of biological pathway models, that may be developed and maintained by teams distributed globally. This architecture computes solutions in a parallel manner while offering ease of maintenance of the integrated model. The individual biological pathway models can be represented in SBML, CellML or in any number of formats. The EGFR model of Kholodenko with known solutions is first tested within the CytoSolve framework to prove it viability. Success of the EGFR test is followed with the development of an integrative model of interferon (IFN) response to virus infection using the CytoSolve platform. The resulting integrated model of IFN yields accurate results based on comparison with previously published in vitro and in vivo studies. A open web-based environment for collaborative testing and continued development is now underway and available on www.cytosolve.com. As more biological pathway models develop in a disparate and decentralized manner, this architecture offers a unique platform for collaborative systems biology, to build large-scale integrative models of cellular function, and eventually one day model the whole cell.

Epilogue

In the Summer of 1978, I was 14 years old and had just completed my sophomore year at Livingston High School in New Jersey. That summer I had been fortunate to have been accepted into the Courant Institute of Mathematical Sciences gifted students program in computer science at New York University. During that Summer, I and 40 other of my fellow students learned eight different computer programming languages including FORTRAN, COBOL and PL/1 to name a few. The course that summer at NYU, and my deep interest in mathematics, in retrospect were important elements that supported my future work.

Swamy Laxminarayana, or Swamy as I called him, however, was my gateway into a completely new world from which I can trace nearly all of my accomplishments. Swamy exposed me to the world of pattern analysis and recognition at an early age. He was a friend above all; although I was nearly 30 years younger than him, he treated me as a colleague with great respect, genuine warmth, and love. My mom, Meena Ayyadurai, worked at the University of Medicine and Dentistry of New Jersey (UMDNJ) as a Systems Analyst while Swamy worked upstairs with Dr. Leslie P. Michelson in the Laboratory for Computer Science. My mom had first introduced me to Dr. Michelson with whom I had begun to build an electronic mail system. In fact, that electronic mail system was the worlds first E-Mail System, for which I received recognition by the Westinghouse Science Award in 1981.

I recall while I was in the midst of building that E-Mail System in 1980, Dr. Michelson began interviewing candidates to find a research staff member to do new research in biomedical engineering. I remember Swamy coming to see Dr. Michelson for the interview. He was dressed in a brown suit with a tie and beige shirt. His hair was worn back with light streaks of grey, he had a peppered mustache, and greeted me as we passed with a huge smile. He appeared very eager, with a mission and purpose.

Swamy, like my mom and I, was one of the few Indians who worked at UMDMJ. Since the chance meeting in Dr. Michelson’s laboratory, I saw him again at various times passing in the hallway. Again the greetings were silent with simple exchanges of smiles. One day when I went home for dinner, I saw Swami seated at our dinner table! My mom, always generous and kind, had invited Swamy over to our house for dinner as she had bumped into him in the hallway. After dinner, Swamy and I sat down and spoke for the first time. He seemed to be on a mission, in a gentle way, to convince me to pursue science, and use my skills in science, mathematics and computer science to help the world.

In particular that conversation opened me to the world of pattern analysis and pattern recognition. I remember that conversation well. Swamy began by telling me about some research he and his sister had done in India. He used this research example to give me an example of how pattern analysis could be used. The example was rather unconventional compared to his conventional research assignments. He and his sister were very curious about seeing if there was a correlation or pattern to peoples palm prints and the onset of disease. In short, Swamy was interested in exploring if there was any truth in palmistry. He explained in detail how they had collected nearly 1000 palm prints from various people. They then created a database with each individual’s name and their palm print. The palm print was a hand drawn sketch. They then had asked each person for their health history including any major diseases they had had or were suffering from. This health history was attached to the palm print.

Each palm print and the associated health history were reviewed manually to see if there were correlations between palm print features to health history. According to Swamy they had found some clear patterns. Certain palm print features had a high correlation to certain forms of disease. He provided me various examples of what he and his sister had discovered. Here was mathematics and modern science being applied to understand an ancient practice. Palmistry itself was a method of visual pattern recognition. Swamy’s example intrigued me.

Swamy then provided me another example of how pattern recognition could be used. One of Swamy’s passions was to model the human heart’s electrophysiological behavior. He shared with me diagrams written on notepads at our dinning room table of the then current theories of how heart signals propagate. It seemed quite complex. He advised me to explore this field since there were many unknowns which would require someone good at both mathematics and computer science to model such phenomenon. Remember, this was at a time when the use of computers was just in their nascent state in the biological sciences.

That dinner conversation really got me thinking. While I had learned a lot of mathematics, was a star student in my high school, and was challenging myself in building the E-Mail System, something in my heart awoke. Swamy’s conversation inspired me to explore how the skills I was developing could be applied to pattern analysis. Moreover, applying computing to biology seemed fascinating. Over the next several months, Swamy and I kept in touch and would have lunch together in the UMDMJ cafeteria. He was always encouraging of my work and kept asking me to work with him. In the Summer of 1981, I was close to completing a version of the first E-Mail System and would have more free time. I had just been accepted to M.I.T., and I promised Swamy to do a project with him, even if it was long distance from M.I.T.

During that first semester at M.I.T., the Institute had recreantly implemented UROP, Undergraduate Research Opportunity Program. One of the research opportunities was to work in pattern analysis for Tadoma. Tadoma is a method by which deaf-blind people communicate. Few understood how deaf-blind people were able to “listen” to someone else through the tactile method of putting their hand on someone’s face. I mentioned this project to Swamy in a phone call and he encouraged me to pursue it since it would provide me some hands-on learning on pattern analysis project.

I kept Swamy posted on my activities at M.I.T. In one of our conversations, Swamy told me about another interesting project that involved heart electrophysiology relative to young infants. Apparently there was a phenomenon where young infants were dying in there sleep. This was called Sudden Infant Death Syndrome (SIDS). In SIDS, babies died in their sleep from what was known as an apnea. Today many hospitals and sleep labs test people for sleep apnea. At that time in the early 1980s, research in SIDS was just a new field. Swamy, through his collaborations, had acquired access to the best infant sleep data through Montefiore Hospital in New York City. This time series sleep data provided sleep states of thousands of babies as well as points in time at which they had the occurrence of an apnea. Swamy’s thesis was that babies sleep states, or patterns of sleep states, may be indicative of the onset of an apnea.

Swamy gently prodded me to get involved in this project to build algorithms to test his hypothesis. Over the next several months, he provided me various books on Haar and Walsh Transforms as well as books on signals processing. Signals processing was a passion of Swamy’s. He wanted me to learn as much as possible in this field for he knew that it would be a strong foundation for future work in pattern analysis. I as then a 17 year old had no idea of its long term importance. The Tadoma project I was working on at that time provided me experience in data acquisition but not in actually developing algorithms.

Swamy, to support my learning, advised me to write up sample programs using Haar and Walsh Transforms. I took his advice and learned a great deal about these two signals processing methodologies. Next year, Swamy provided me raw data of SIDS from Montefiore Hospital that he had acquired in 1977. He also provided me various papers he had written dealing with SIDS in his previous work. The earlier programming project I had done on Haar and Walsh Transforms was valuable to the SIDS project. In this project, my task was to review the data and see if I could find patterns of sleep states leading to the onset of an apnea.

The SIDS project was like solving a puzzle. Swamy taught me that babies have six states of sleep, whereas adults only have five. The data resembled up and down steps, each step being a sleep state. There were certain points which were marked when an apnea took place. The goal was to look backward from the point of the apnea to see if certain patterns sleep state correlated to when an apnea occurred. I used the earlier Haar and Walsh Transforms to build pattern analysis methods to predict the onset of an apnea from the sleep waiting times.

While traveling back and forth between M.I.T. and UMDNJ, Swamy never micro managed me but was always there with patience to answer any of my questions. I worked on this for over a year. Some valuable results came from the analysis. The next year, the Fall of 1983, we had heard about the international conference to be held in Espoo, Finland. This conference on medical and biological engineering was the worlds largest such event. In the world of academia were as someone once said, “people fight over nothing”, Swamy was unbelievably generous, again in retrospect. He wanted to include my results in a paper on SIDS, and where many in academia would have never thought about giving an undergraduate authorship on a paper, Swamy made me a co-author. In addition he invited me to come to Espoo to co-present the paper with him.

Going to Finland would be my first trip out of the United States to a foreign country besides India. Our paper was accepted for the Summer 1984 Conference. I was on my way to Finland. Attending that medical conference was an amazing experience. Twenty-seven years later today, I still remember that trip vividly. I was the youngest registering at the conference, but Swamy introduced me as his colleague from M.I.T. to all of the other scientists. For a 19-year old to go on a flight, attend banquets, hear scientific talks, travel the Finnish countryside, and hear a new language, was out of this world. Swamy had made that experience possible for me. He exposed me to a larger world, beyond science.

It was clear to me that scientists went to these conferences not just for the science but to experience other lands, local cuisines, and meet other people. Had Swamy not taken me on this trip, I would never have been exposed to this aspect of science. After coming back from that Conference, my enthusiasm for science and pursuing research in pattern analysis and pattern recognition skyrocketed.

During 1985–1994, to the beginning of my Ph.D. at M.I.T., I participated in numerous pattern recognition research projects. Swamy and I would talk from time to time on the phone, and I would keep him abreast on my work at MIT. He was always so very supportive, encouraging and uplifting. He was always positive and always ready to help. Swamy shared with me only a bit of his personal history. All I knew is that he had been in Holland for sometime, had gotten married to a Dutch woman, and had children. His moral support of my research activities helped me in my research pursuits. Among the projects I was involved in included: handwriting recognition, ultrasonic wave analysis, document analysis, image flow visualization, and a number of other pattern analysis projects.

From these various pattern analysis projects, there appeared to be a common set of strategies and methods that could be abstracted across all fields. That thought led to my Ph.D. thesis at M.I.T. entitled Information Cybernetics. In 1993, while in the middle of my Ph.D. work, I was invited to participate in a very interesting pattern analysis competition. This competition was not scientific in origin; it was from the U.S. Government. In particular, the Executive Office of the White House, with then President Bill Clinton, was looking for intelligent ways using computers to sort their E-Mail. At that time, the White House was receiving nearly 5,000 E-Mail per day, and this was before the introduction of the World Wide Web. There were 147 different categories. Student interns at the White House were manually reading each E-Mail and assigning them into one of the 147 categories. Categories included drugs, education, death threats, etc. The White House was interested in automatic filtering and sorting of E-Mail for two reasons.

I approached the White House competition with little knowledge natural language processing. My approach was engineering, where I used a hybrid method of employing nearly 19 different methods spanning feature extraction, clustering, to supervised and unsupervised learning. I was the only graduate student involved. The other five competitors were private and publicly traded companies. I won the competition.

I took time off in M.I.T. in 1994 to start EchoMail, a company for pattern analysis of electronic mail. We grew the company to nearly 300 employees worldwide by year 2000. I remember hearing from Swamy once during that time asking me for a letter of recommendation as he was taking a new job in the Midwest. It was weird for me to write a letter of recommendation for him. He had written several for me many decades before. Following that interaction I had not heard from him. It was my mom who a few years later informed me of his passing.

Hearing of Swamy’s passing was sad. He had always been there and now was no more. I wondered who was with him, what his life had been during those past 10 years when I had lost touch. I recall Swamy and I sharing at our first dinner conversation thoughts about ancient Indian science, its mystical and scientific aspects, including the concept of Soul. There was an unstated agreement that Soul never dies. As I heard about Swamy’s passing, I flashed back to that dinner conversation and knew his spirit would and had never died.

As I look back from the time of my meeting Swamy till today, 2010, my entire history of work, particularly my success with EchoMail, development of CytoSolve, and my recent research at M.I.T. to discover patterns of coherence that bridge traditional systems of medicine with modern systems biology, can all be traced to that dinner conversation with Swamy. As I write this last sentence, I am deeply moved and wish Swamy was here so we could have dinner again and I could thank him for the wondrous gift he gave a 16-year old nearly 30 years ago.