Fourth Week

Hello lovely readers,

Week 4 has flown by, and thus a new blogpost is required. This week was a short one for me, as I spent Monday and Tuesday at home.

It felt good to be back home for a couple days. I wasn’t feeling homesick or anything, though. Here in Leiden I have to pay attention to how I speak so people can actually understand me, so it felt great to be able to speak like I normally would. The Flemish accents all around were a breath of fresh air after a bit more than 3 weeks of Dutch.

I returned to Leiden Tuesday evening, stepping off the train around 12pm. The very next day I went back to my internship a bit tired from the long-ish train journey.

In terms of work, I have added more download buttons, so all datatables can be downloaded in two formats. There are still some kinks to be hammered out, but they do work. So I think I’ll focus on the more important aspects of my application. One of which was the calculation of several values, which, as it turns out, was not correct. I went through my code to edit some parts, and I noticed I had not changed some of the objects in the calculations for one of the options. As such, it used the objects of another option to calculate values, and produced faulty results. Trying to fix this made me realize that the way I was doing it now wasn’t going to work.

I’ve spent a lot of time searching the internet for ways to fix my code and calculations, but couldn’t really find anything concrete I could use. I browsed sample workflows to find solutions, but none of them had something convenient I could just put into my code to make it work. However, there were some bits and pieces scattered around, and if I could combine them, I would get my desired result.

Thus, I read the workflow, entered the code in a different file, and ran it with my data to see what exactly what effect the different lines had. Everything was going pretty well, until I hit a roadblock. The code wanted to calculate the dispersion of my samples, but my data had none. The code needed dispersion to execute further functions. I didn’t know what to do, and went back to Prof. Dr. G. Oogle for advice. Sadly, nothing I thought might help worked, and I was out of ideas. So I reread my code another three times or so, comparing the sample workflow with my own.

And then I saw it. The problem. A stupid mistake. I had been following the workflow too closely, not completely understanding how they had organized their samples, and had messed up with my own samples. As such, I had ended up with 16 groups (1 for each sample) instead of 4, and eliminated the dispersion by accident. I quickly fixed the issue, and lo and behold! I had data to work with.

The difficult part is also the part where this week’s work ends. With the workflow I could calculate the logFC, the logCPM and the Pvalue for a sample, but I still needed the contrasts and the meanCPM and the feature names from my previous block of code. I was fairly sure I could write all of the parameters into a table with the help of some loop structures, but it would take a while to figure out. Last I managed before the end of the week was using the loops to create half of the datatable I wanted.

The weekend felt short, but I had some fun. I went to a jazz café Friday evening. The beer was expensive (granted, it was craft beer, but in Belgium a special beer is still cheaper than what I paid here) though I did like the taste. The music, of course, was excellent. Saturday was a lazy day. I went swimming in the morning. I thought I would feel more awake afterwards, but when I came home and took a shower, I felt more tired than I had been before. I didn’t end up doing a whole lot afterwards. I did slog through a bit of administration I’d been neglecting (and overwrote one of the documents, so I had to fill it out again), so that’s at least something. On Sunday I went to the local game store to play Adventurers League during the afternoon, and created a new jank deck on MTG arena. I don’t think it’s viable, but it is fun to pull off the combo.

Anyways, that’s it for this week. I’ll see you in the next post!

Tibbert