Just as what presented in the title, I finally finished my master dissertation last day. When I look back upon my whole three-year master career, I realized how much of my progress and deficiency.
Technically, I have practiced myself into a Swiss Army Knife. I am absolutely a bioinformatics novice when I came to the Academy of Military Medical Sciences. I was a wet lab student before, which means that I know how to prepare a solution more than how to write a snippet of code in a black screen. When I began my graduate life, the main challenge is domain disparity. What my supervisor gives to me is an absolute text mining topic — short text classification. Frankly speaking, I am not much into it, but as a junior graduate student, I know nothing knowledge about negotiating. Therefore, I keep reading mountains CS literature and coding in python (thank Guido). Python is an interpreted language that is helped me a lot in the transition. It’s a total shift for me, I love change, but I do not love the change to that topic none of biology at that time. After almost 6 months grind, I surprised found my coding skill improves a lot and I talked to my supervisor to apply text mining in a more specific way — finding new prebiotics by text mining technology. He was positive because, at that time, our lab was obsessed with finding new prebiotics. Thus, we got a win-win strategy and my coding skill trained by short text classification can have its own place in mining new prebiotics, it’s really a big step for me.
After the text mining work, I was considered to do some bioinformatics analysis, such as next generation sequencing (NGS) or cancer genomics. Unfortunately, my supervisor told me he wants to develop a WeChat platform. WeChat is the most popular instant messaging app in China, and its public platform can allow developers to customize their own needs through programming. So, I developed a PubMed Customized Retrieving system (PCR) on WeChat Public Platform. The service provides several convenient features such as literature customization, literature querying and Impact Factor (IF) Querying by parsing input, matching built-in journals name table dynamically and retrieving PubMed database [WeChat paper]. This work helps to improve the literature learning efficiency and reduce the time cost for researchers. At last, almost at the half of my graduate life, I can finally do some real bioinformatics work. I dive into The Cancer Genome Atlas (TCGA) and began to cancer genomics research. After one year work, I do explore the network structure of genes and miRNAs at the system level and to study topological characteristics of survival rate related molecules by means of survival analysis. It turns out that ‘hub’ nodes in gene regulatory networks promise to be potential features for molecular subtyping [Network paper]. When finished the paper, I began to study the relationship between mutation and cancer chemotherapy resistance to find biomarkers, at last, we found a set of mutation biomarkers which can predict chemotherapy resistance by three machine learning technologies (RF, SVM, LASSO). It is worth mentioning that the AUC (Area Under the Curve) in HNSC (Head and Neck Squamous Cell) can be up to 0.980, which is really a high performance.
In a word, these three years master life gave me a lot — Skills, friendship, and love. But there is still existing shortcoming I need to pay attention in the future. First, not hesitate to communicate with your boss and reach a win-win strategy can make both happy. Second, trying to make consecutive research in your career, which cannot only help you graduate but also build your academic reputation. Last but not the least, no matter under what circumstances, be positive.