Expectations. First week of the program, I had in mind three things I wish to accomplish: to learn how to ask questions as it’s the important first step of research, to reach out and take initiatives such that we would be prepared for the “all-for-yourself end of continuum”, and most importantly, to be consistently happy and persistently passionate even when the outcomes fail me. Now six weeks, which is a long time, has passed—eight hours a day and five days a week—I got from from knowing nothing about UNIX command lines or R to having a fair amount of knowledge about shell programming and statistics as well as the biology behind it. However considering the big picture, eight weeks is also really inadequate in the scope of a research project.
My goal seems to be simple enough—validating the statistical model my mentor previously developed and analyzing the results we get, which I thought would not take the entire eight weeks to finish. However, after spending weeks reading literature and learning to program and to use the model, when I finally ran the codes to visualize TF binding across different cell types by heat-mapping, we got some unexpected results (which seemed to be the theme of the week with the World Cup…). The predictions of the models and the reference (ChIP-Seq) sometimes don’t match well, with transcription factors known to bind in a ubiquitous manner being predicted more effectively than cell-type specific ones. This was unexpected since the model is trained for each cell type independently and the predictions should be able to reflect the difference. This is really exciting for me because it means there’s some hidden stories to be discovered! My mentor and I hypothesized that the reference data itself is not entirely accurate, as they are obtained from different procedures and peak-calling protocols. So for the next week, I’ll look at these sites on genome browser as well as run the models on a different transcription factor (REST) to see how the prediction performs. And for the first time, I’m loving the unexpected (not that I hope the model doesn’t work, but that having unexpected results means there’s something we are not aware of behind the scene).
Six weeks is long enough for me to have a taste about what research is like, but short enough to keep me passionate about what im doing and prevent the frustration or tedious work from giving me second thoughts. So at this point, I still don’t know if research is for me; but I enjoyed and am loving every day of the process.
