So ... is my analysis correct?
One day has passed since submitting the research article, and yet I am still paralyzed with fatigue from typing up the whole article. How much time did I spend, I wonder, sitting in front of my computer making figures, writing up the article, rewriting them again and again. Every time I try to count the hours spent on it, the haze in my memory hinders me from giving the exact duration; I am simply left with the fact that it was way over 20 hours.
Sure, I knew that the research article would take longer than the data summary in the Module 1. It is no longer a group project and every section has to be written in full sentences. But the increase in the work load cannot be explained by those trivial factors. The cruelty of the workload stemmed from the data that we were handling, a completely different kind from Module 1.
Module 1 had a simple story in the experiments we conducted. You want to find a new binder for FKBP12, and you test it. What is great about this story is that you are guaranteed to be doing something new. On the other hand, experiments in Module 2 was i) hard to synthesize into one cohesive story, and ii) can be a simple replication of past experiments. Just saying that etoposide treatment induces stress response and cell cycle arrest is of no use, since this is a phenomenon that is already observed; in other words, no knowledge gap is filled with such observation. Hence our task was to find something new from the massive data set that was given. And this is indeed the part I struggled the most.
My article turned out to be computational-heavy, as most figures and analysis came from RNA-seq analysis on R. In order to find something new, I tried out many analyses. After series of unfruitful trials, I finally found something interesting that can be tied to the density-dependent sensitivity of BRCA2 that we found in the cell viability assay. But then I started worrying. Is my analysis actually correct. If I was coding for a elementary CS course, I can just run my code on test cases to see if works. But in this case, how can I know that my code is analyzing the way I intended to.
The answer to this, I still do not have. I made sure the code made sense in my head, and summarized the analysis in the paper, but there is still a chance that there is an error in the analysis. In real science, I guess this is why each scientific findings are confirmed by multiple research groups to be established as a ground truth. I learned this lesson through the agony and pain.
Sure, I knew that the research article would take longer than the data summary in the Module 1. It is no longer a group project and every section has to be written in full sentences. But the increase in the work load cannot be explained by those trivial factors. The cruelty of the workload stemmed from the data that we were handling, a completely different kind from Module 1.
Module 1 had a simple story in the experiments we conducted. You want to find a new binder for FKBP12, and you test it. What is great about this story is that you are guaranteed to be doing something new. On the other hand, experiments in Module 2 was i) hard to synthesize into one cohesive story, and ii) can be a simple replication of past experiments. Just saying that etoposide treatment induces stress response and cell cycle arrest is of no use, since this is a phenomenon that is already observed; in other words, no knowledge gap is filled with such observation. Hence our task was to find something new from the massive data set that was given. And this is indeed the part I struggled the most.
My article turned out to be computational-heavy, as most figures and analysis came from RNA-seq analysis on R. In order to find something new, I tried out many analyses. After series of unfruitful trials, I finally found something interesting that can be tied to the density-dependent sensitivity of BRCA2 that we found in the cell viability assay. But then I started worrying. Is my analysis actually correct. If I was coding for a elementary CS course, I can just run my code on test cases to see if works. But in this case, how can I know that my code is analyzing the way I intended to.
The answer to this, I still do not have. I made sure the code made sense in my head, and summarized the analysis in the paper, but there is still a chance that there is an error in the analysis. In real science, I guess this is why each scientific findings are confirmed by multiple research groups to be established as a ground truth. I learned this lesson through the agony and pain.
Comments
Post a Comment