ShapBio Lab Notes

Maybe this time

Josh — Mon, 04 Oct 2027 00:00:00 GMT

I really did think two years ago that I was going to get the website up and running. I really did. I think I even got kind of close, but not close enough. But now I think it might really happen, so I’m writing another post¹.

¹ I think part of the reason I am motivated again is that for the first time in a while I have my own personal computer that I am explicitly not using for work. This has to do with some changes in policies at work, but also the fact that I could get a cute little MacBook Neo, and dang if it isn’t cute and fun. Blue, if you were wondering.

If you look at the code here, you will see that this time around I got some help with rewriting the code using Claude. Specifically, I got it to help with the long-planned and long-delayed transition from Jekyll to Quarto.

What I did not do (and specifically banned in my AGENTS.md file) is any AI assistance with writing content. Everything you see as content is written by me, with some possible use of autocomplete suggestions within code blocks.

I can imagine a time when I may include some more AI-written code sections, but honestly I have some doubts about the extent to which I will include code much at all. I had some plans to do that a couple of years ago (see my post about path stripping), but writing code is usually well handled by LLMs, so it seems a bit pointless these days. On the other hand, the code I see the LLMs writing is often not really well done, so maybe more good examples are worth the effort. Not that my code is always good! I just had a look back through the code I had written for my biostats class and let’s just say I wouldn’t teach with that code these days. So we’ll see.

I’m kind of hopeful that I will be inspired to write some more, but I have been hopeful before.

Session info and security through obscurity

Josh — Mon, 03 Mar 2025 00:00:00 GMT

In my work at the Data Lab, I have become accustomed to always including session information at the end of a computational notebook. It is kind a minimal step toward reproducibility to record what versions of all of the R packages were used to generate the output.¹

¹ We also use renv as much as possible to make it easier to reproduce the environment we are using, but it is nice to have both.

The {session_info} package makes some of the formatting and information provided nicer, so I try to use that most places, including on this blog, though I will happily fall back to the built in sessionInfo() function if I don’t want to add dependencies.

That said, there is one thing that does annoy me, which is that the paths that are printed are full paths. I get it, but I don’t particularly like it, because it seems a minor breach of best security practices to be printing full paths on your computer all over the place. I mean, we all do it, but it doesn’t mean I like it. I also don’t think this information is particularly useful to anyone, so it seems kind of silly to include it.

For this blog then, I wanted to strip out as much of that info as possible. Since I use {renv} most of the time, I expect most paths to be nestled somewhere under the path given by renv::paths$root(). Anything else in the project directory will be somewhere inside the path defined by here::here() (the core function of the wonderful {here} package). So I start by stripping those out, as well as anything under a common home path, replacing the values with the environment variable that they would use.

Here is the function I wrote to do that²:

² The last str_replace() strips out anything in /User or /home subdirectories, which may not actually be the same as ${HOME}, but it is likely somebody’s home. So that seemed reasonable to me.

strip_paths <- function(path) {
  # replace user paths with variables
  path |>
    stringr::str_replace(
      stringr::fixed(renv::paths$root()),
      "${RENV_PATHS_ROOT}"
    ) |>
    stringr::str_replace(
      stringr::fixed(here::here()),
      "${PROJECT_ROOT}"
    ) |>
    stringr::str_replace(
      stringr::fixed(paste0(fs::path_home(), "/")),
      "${HOME}/"
    ) |>
    stringr::str_replace(
      stringr::regex(
        "/(Users|home)/[^/]+/",
        ignore_case = TRUE
      ),
      "${HOME}/"
    )
}

I then use that function to relabel the factors in the $packages$library slot of the object created by sessioninfo::session_info(). Print that out instead of the original object, and there we are.

You can see the full code and results below. It’s not perfect, but it seems to do what I want, at least for now.

R session information

It’s only been two years

Josh — Sun, 02 Mar 2025 00:00:00 GMT

Right, so it has been a bit over two years since I said I was starting to write this blog. Obviously, I didn’t get too far.

Well, that isn’t fair to myself, as I have done many things in the time since I started reformatting my website. I just haven’t gotten around to actually writing more, or finishing things off. At least I knew then that I would only be posting irregularly.

Today I updated a bunch of the formatting, and even updated the CV page with publications from after I started at the Data Lab.

I might even push all of these changes and have the site go live this week. Might.

I do want to get this up, warts and all, to at least have something up to correct the embarrassing situation that there is still a website up that claims I hold a position I haven’t held for 5 years at this point.

R session information

Let’s see how this goes

Josh — Sun, 01 Jan 2023 00:00:00 GMT

I’m waiting on the new year as I write this. I probably won’t finish this post or publish it for a while yet, but it seemed a good time to start. My personal website is horribly out of date, I had a bit of free time while visiting the in-laws, and there is a not-so-new literate programming engine on the block, so I thought I would try to get things going.

I’ve never had much luck with blogging. I usually don’t feel like I have much to say, and the effort to get posts up is usually beyond my motivation, so the blogs I’ve started in the past have never gotten more than a couple of posts. If this one does better, well, fantastic. We’ll see how we go.

Just a couple of programming notes as we get started:

I’m rebuilding my website with Quarto, which is the literate programming engine mentioned above. As an evolution of Rmarkdown, I’m pretty familiar with the basics, but not as much the full website features. Previously, I had used Jekyll, and had a fair amount of customization that I had built up in there… Most of that is being ripped out for now. Maybe I will customize more in the future, but as I age I have gotten less and less interested in tweaking things to be just so, as I usually regret it later. So, back to basics, sort of.
I expect most of my posts to include some code, mostly R, but probably some Python and Nextflow, since I’ve been spending chunks of time with each of those. I have no idea how long a post will usually be. Probably pretty short. If I set that expectation for myself, I might actually write some things. One thought is to try to use this blog as a place to make some notes as I discover tips and tricks that I want to record for myself, and others might find useful. I’ll probably complain some about things that have frustrated me, but I’ll try to be nice about it. And maybe people will read this and give tips.
Most posts will include R session information. Why? Partly because it is a habit I have gotten into in my work at the Childhood Cancer Data Lab, but partly because it seems like a good thing to do. If there is code in the blog post, I want to be able to see what version of the various packages I was working with when I come back years later. When rebuilding this site, I found a number of places where my code was old and broken. Some of those I updated, some I just deleted. Quarto makes it fairly easy to freeze results as they were produced, but I still want that record for future reference. It’s neat to see how code and conventions evolve. This is probably a subject for a future post, if I can arrange my thoughts. But for now, let’s just see what we were starting with as we got started with this new version of the site¹:

¹ Update from the future: this was not what we started with, as it took me a while to get everything done, and I wanted to change some formatting nonsense, and that meant rerendering things, including this session info.

R session information