Pinned toot

I'm a PhD candidate in Medical Biophysics at the University of Toronto studying epigenetics and cancer. Math and physics background taking a step into bioinformatics and evolution

covid, reinfection

Again, overall these are optimistic results. Let's hope these findings generalize to humans as well.

That would be good news for all of us

covid, reinfection

All 3 groups showed a reduced viral load after re-infection. It was still present in the high and medium exposure groups 1 day after re-infection, but not detectable in the low-exposure group.

Viral load was not detected after 6 days in all 3 exposure groups. These levels were all much higher in the initial infection

covid, reinfection

The second paper performed a similar experiment. They had 3 groups with different levels of initial viral exposure (group 1 = high infection concentration, group 2 = medium, group 3 = low). Viral load was detected after infection for each group

covid, reinfection

Viral load after initial infection was found in various tissues. However, in the re-infected group, viral load was not detected in any of the same tissues.

This suggests the infection and re-infection were both done properly and that the virus was cleared and no longer present, even after re-infection

covid, reinfection

From the first paper, you can see changes to body weight and temperature after each infection.

All subjects showed increased viral load after initial infection. The group that was re-infected did not show an increased viral load after re-infection.

The group that was re-infected did show a change in immune cell counts after re-infection whereas the monkeys that weren't re-infected showed no changes in the same time

covid, reinfection

Two papers testing reinfection with SARS-CoV-2 in rhesus macacques. They both show minimal viral load after reinfection.
This suggests that reinfection may not be a concern.

Obligatory "this is not in humans" and "I am not a virologist" and "these are small sample sizes". But these are optimistic preliminary results, for sure

If this isn't the best answer to "what makes research hard?", please point me towards it

I'm now a faculty associate or whatever at UCL's Center for United States Politics.

It literally started with me congratulating my office mate on getting it off the ground, and him responding, 'hey, wanna be a part of it?'.

I wanna borrow his playbook and create a UCL Center for Things That Interest Me. It may be a bit eclectic, but it'll be interesting (at least for me)!

This is a great flowchart on thinking about data-driven stories, and scientific thinking more generally

pudding.cool/process/pivot-con

Another excellent article from The Pudding

It's been a while since I've done some mathematical physics, this was the type of low level intuitive stuff I always liked learning about. How fundamentals of scientific theories are reflected in the models you use to study it is always interesting

The second is about how symmetries and conservation laws are fundamental to physics if only because that's what we want physics to look like

jrhawley.github.io/2020/08/03/

This one was fun, I was really inspired by some recent posts by John Baez and took the time to learn about abstract algebras, what they are mathematically, and some ideas about how observables and generators are represented by objects in math

Two new blog posts out!

The Canadian Civic Holiday gave me some time to put the finishing touches on some ideas and read some other blogs on the subjects I was thinking about.

The first is about the risky footing scientific research is going on if it doesn't get serious about good software development practices

jrhawley.github.io/2020/08/03/

Fediverse app and blocking bad domains

Whenever someone mentions @fedilab or similar apps which chose not to implement a domain block of gab etc., people demean it and call the app "fascist". I stand by this statement:

f-droid.org/en/2019/07/16/stat

My personal take, implementing a block list is a full time job. Look at how adblock addons rely on a number of community maintained lists of advertising and tracking domains.

Another committee meeting done! Hopefully the last one before I ask for permission to write and graduate. Let's see how the next since months go

It's certainly not perfect, but at least I find that it separates those 5 points in a way that makes them clearly separate things.

I never understood why the "sample standard deviation" was calculated with 1/(n-1), while the actual definition for the standard deviation used 1/n. Turns out it was all because I didn't get the difference between the random variable that is the sample standard deviation and the distribution from which it came

So we observe a single value $x$ of a random variable $X$ whose distribution is $\mathcal{X}$.
If we make multiple observations of separate individuals, then we would have observations $x^(i)$ of the random variables $X^(i)$, each of which are distributed as $\mathcal{X}$.

The observed mean would be $\bar{x}$, which is an observation of the random variable $\bar{X}$, whose distribution is $N(\mathbb{E}[X], \mathbb{V}[X])$, by the Central Limit Theorem.

In LaTeX, I generally try to use these notations for the above categories:

1. \mathcal
2. uppercase letters
3. lowercase letters
4. lowercase letters with superscript (i) to denote the i-th observation (leaving subscripts for elements of vector-/matrix-valued observations)
5. \mathbb

Overloading the same variable doesn't make a good inverse mapping from the abstract objects back to the expressions, and makes it hard for people to know how to handle those abstract objects

In statistics, I have never found a good notation that matches these descriptions that covers

1. Distributions
2. Random variables
3. Observations of random variables
4. IID observations of a group of random variables
5. Operators on a single or group of random variables

I've found variable names are consistently overloaded for 2 or more of those categories in texts, which, imo, leads to the confusion of those ideas in people's minds