SAMSI-SAVI Workshop on Statistical Methods for Bioinformatics: December 2013

The following was written by Malay Bhattacharyya, Department of C.S.E., University of Kalyani, India

Malay Bhattacharyya

Malay Bhattacharyya

“Don’t give a talk, take a class.” This was the driving force, as I feel, behind the SAMSI Workshop on Statistical Methods for Bioinformatics we had at the IISc campus, India during Dec 12-14, 2013. It included people from diverse backgrounds encompassing biologists, statisticians, mathematicians, computer scientists, biostatisticians, biophysicians, biochemists, anthropologists, and lot more.

The Department of Mathematics at IISc was a perfect inspiring venue for this workshop, particularly for research discussions. It has chalks and boards kept everywhere, at every corner of the department! Even I saw a catering person to write random equations (although trivial) on a board avoiding others’ notice. Environment really motivates!

The very first talk of the workshop, by Varghese George of Georgia Regents University, made us feel like entering into the revitalized world of the epigenetics. The recent progresses and futuristic goals were very nicely introduced. Both Indranil Mukhopadhyay of ISI Kolkata and S. R. Deshmukh of University of Pune had comprehensive introductory talks afterwards about the basics of statistical tests, molecular biology, etc. and about expression profiling, respectively. Many of the speakers were also benefited from their efforts of making a strong foundation of the preliminary concepts. It was not required for them to start from the scratch.

The first day was so windy that I felt possibly the air is also keen to enter into the lecture hall, a perfect learning platform! What I liked most is not only the respective speakers were responding to the questions, rather everybody took the pain to discuss and settle the best answer for the tricky ones. The sessions were not too much attendee-heavy, so everybody had a fair chance of asking questions. Again, the ratio close to 2:3 between the invited speakers and the participants set up a real platform of face-to-face learning.

Naomi Altman of Pennsylvania State University gave an attractive talk on the recent obesity of the high throughput data. The reproducibility of research became a major issue of discussion lead by the talk of Prof. Altman. It is really a burning issue worldwide. There was a common agreement between all the invited speakers, who strongly encouraged keeping the relevant source codes also available alongside the publications.

How statistical analysis can help in some particular areas of plant genetics, especially in the gene duplication problems, was thoroughly described by the SAMSI Deputy Director Snehlata Huzurbazar. It was also a real pleasure to have Ashis Sen Gupta of ISI Kolkata on stage to talk about his work on understanding the circadian rhythms. It is a real challenge to build statistical models given the surprising fact that “the rhythm of the life is circular, although the life is itself a linear game.”

The next day we also had a great experience! Everybody was cracking jokes with the “Friday the 13th” issue that eventually marked out the second day of the workshop. But it was really an enjoyable day – not only for the banquet but also for the diversity and depth of the talks. It started with the talk by Nagasuma R. Chandra from the host institute who gave a realistic overview of how to proceed step-by-step towards the modelling of disease prediction. She has a strong belief that making such systems automated will indeed speed up the progresses in this direction.

Olga Vitek from Purdue University gave us an all-inclusive glance toward the immense scope lying in some promising areas of proteomics. Again, N. Srinivasan of IISc detailed on a fantastic account of his research on finding missing links between protein families using computational models.

T. S. Vasulu of ISI Kolkata and Paul Joyce (I love the way he explains complex things with funny examples) of University of Idaho gave nice introductions to statistical methods that can be applied to phylogenetic analysis and for the study of adaptive evolution, respectively. Switching between the laser pointer and the hard pointing stick was a real fun for Prof. Joyce.


Every talk was made somewhat flexible based on the demand of the attendees. I remember a talk to have been stretched by half an hour or so to satisfactorily answer every question raised by the listeners. Still the overall time frame for the entire day was well maintained. We also had a nice photo session on the terrace of the hosting department in a mood of get together.

The long walks (voluntarily avoiding cars) through the woods of the campus of IISc with many of the speakers, while returning after the workshop days, gave me pleasant chance of gaining additional experience through informal discussions. Research is no more an independent effort, rather a collaborative competition.

The final day of the workshop started with a nice talk by Susan Holmes of Stanford University, who highlighted diverse facets of the Human Microbiome Project. Her idea of more on leaps less on slides was great, using the chalks and boards every time. The concluding talk by Sanghamitra Bandyopadyay of ISI Kolkata, a very basic one bridging between statistics, computer science and biology, detailed on various robust computational models that can tackle multi-objective problems, often occurring in expression analysis and related areas.

The contributory talks by seven of us, mostly covering ongoing studies, were strongly benefitted from the expert speakers being also a part of the audience. I enjoyed my contributed talk and received a couple of valuable suggestions. The take home message was “do whatever you wish to do, but with a clear conception and full confidence.” We were unfortunate this time to miss the talk of K. Thangaraj of CCMB, on the very last day because of his absence due to some urgent involvements.

I hope that the slides of the talks will soon be available online. The logistics managed by Shruti and Sai were fabulous. The foods were so diverse that I experienced the taste of the entire India. The organizers took a lot of pain to arrange a great banquet dinner on the second day. The round table discussions in the banquet session were really effective.

I feel like SAMSI workshop proved to be a real SAMSI (such a mega statistical incident). Hats off SAMSI!

October 2013 Undergraduate Workshop

The following was written by Theresa Gebert, an undergraduate student at Harvard University.

group shot

Undergraduate workshop participants.

As part of its Education and Outreach Program this year, SAMSI offered a two-day undergraduate workshop on “topics of current interest in statistics and applied mathematics.” In addition to an overview SAMSI Research Programs, the program topic was Computational Methods in Social Sciences.

Just like many of the other program participants, I had heard about this workshop from my advisor, Joe Blitzstein, the co-director of the Undergraduate Statistics Program at Harvard. Despite my summer research experience in the statistical analysis of human behavior, I was not quite sure what to expect when I flew from Boston, Massachusetts to Raleigh, North Carolina on a Wednesday night. What sorts of computational methods do social scientists need?

As it turns out: many. These methods range from the simple t-test to complex, high-dimensional network analysis. We learned about the building blocks of social networks from Professor Krista Gile, including what we often exclude in our representations, such as multi-modal networks. We learned about political networks from Bruce Desmarais and the idea that the study of people can actually be a systematic process. We were also introduced to issues of data confidentiality by Jerry Reiter, who explained that recoding variables or hiding values is not always enough. (The solution he suggested, first introduced by statistician Donald Rubin, is fully synthetic datasets!) On the very first day, the undergraduate participants were opened to the world of social science as a quantitative problem, which was certainly an approach very different from the one I had encountered before.

students sitting in lecture room

Students at the workshop.

But beyond the classroom, the workshop also encouraged the social science of learning from others’ experiences as well. Every speaker started by introducing their own academic and life trajectory: what did they start out doing? What do they do now? How have their goals and dreams changed over time? It was fascinating to learn about the power and the limits of our knowledge in the realm of social networks, but it was equally fascinating to hear that people who had studied English, Chemistry, or hated math, ended up finding their way into social science and statistics somehow.

In addition, it was very valuable to get to know and learn from fellow participants as well. I met some of the brightest, most intellectually curious undergraduate students in those two days. Whether we were bonding over the complexity of the lecture we had just heard, the lab tutorial we were trying to solve, or the problem sets we had to finish before we flew back to our respective schools, it was truly an atmosphere of intellectual curiosity and camaraderie. I have remained in touch with several of the participants; I certainly hope I might get the chance to enter mathematics competitions or hackathons with them in the future!

The next day, David Banks presented a fascinating lecture on dynamic network models; starting with concepts directly from graph theory, Banks ended the talk with his own research in the social network of political blogging in the aftermath of the Trayvon Martin incident, which he conducted using statistical techniques in machine learning and language recognition. It was followed by a practical introduction to the software R and a package that enables the visualization of social networks.

Even though the workshop was just two days, it was surprisingly difficult to fly back to Boston after such an inspiring, intellectual hiatus from college life. What was so satisfying about my experience was that I got the chance to expand both my academic and social networks; I had engaging conversations with fellow participants as well as the brilliant SAMSI post-graduate fellows, conversations which never failed to spark new ideas and interests. I got the chance to meet and listen to professors from a variety of fields and high levels of achievement, who were also incredibly approachable and genuine. The workshop completely surpassed every expectation and cemented my belief that graduate school in Statistics is my dream. I certainly hope I will get the chance to become more involved with SAMSI in the future, and I am so grateful to them for making these opportunities possible for the undergraduate community!