Having worked in computational genomic and epigenetic research and collaboration, what I found to be extremely useful is to have a good system to study the function of certain genes or chromatin regulators in the correct context. This reminds me about what Dana Pe’er mentioned lately, “it is all about context”. More specifically, perturbation of the factor of interest needs to produce a strong enough phenotype in a system (e.g. cell line, tissue, tumor), to warrant genomic or epigenomic profiling. Otherwise, additional genomic / epigenomic profiling and bioinformatics modeling might be less cost effective than simply changing the system (i.e. context).
Recently I went to a DFCI faculty training on giving feedbacks, and the instructor gave me a short article titled “We know what they did wrong, but not why: the case for ‘frame-based’ feedback” by Jenny Rudolph and Daniel Raemer. Here is the slides for Dr. Shapiro’s training course, although I found the article to be more informative. It mentioned that actional feedback targeted to the learner’s needs is one of the strongest predictors of improved performance in learning, and proposes a 3-step feedback algorithm:
Step 1: describe the problem from the instructor’s perspective. Tell the trainee unambiguously and specifically what, from the instructor’s personal perspective went wrong (or right).
Step 2: diagnose the frame. Ask questions to discoed what “cognitive frames” drove the trainee’s action.
Step 3: teach to trainee’s frames. Tailor instructions and discussion to the trainee’s frames.
Indeed, how often did I ask students or postdocs why they think the mistake happened and tailor the teaching accordingly? Actually this might work when my son makes mistakes and gets frustrated with violin practice. I will try to give better feedback in the future…
I was checking out Curtis Huttenhower‘s website today, and saw his blog. He mentioned this paper in Nature 2010, where the author talks about the importance of good practices in scientific programming. The articles says good things about python as a scientific programming language. I especially like David Gavaghan’s comment “there needs to be a real shift in mindset away from worrying about how to get published in Nature and towards thinking about how to reward work that will be useful to the wider community” and his “master–apprentice” approach to train graduate students”.
Jun Liu and I are starting to teach the STAT115/215 course starting this week. It is an introduction to Computational Biology and Bioinformatics. Inspired by Eric Lander and Xiaoli Meng, we are trying to improve the teaching. Instead of cramming students with knowledge and maths, we would really like to motivate them to think and enjoy the learning experience. Teaching is the best way for me to learn new material, and we invited several local experts to cover the last module on the frontiers of the field. Thanks to the wonderful teaching fellows, especially Daniel Fernandez, we are revamping the course lecture material, labs, and homework. They also made a publicly available website, with lecture slides and videos. Hopefully it will be a useful resource for others to learn bioinformatics.