Is progeria related to aging in the same way that familial AD is related to sporadic AD?

Attention conservation notice: Someone has probably made this point before.

Progeria is a genetic disorder caused by mutations in the lamin A nuclear lamina protein. Since it manifests in several ways that resemble an aged state (eg wrinkled skin, atherosclerosis, kidney failure, loss of eyesight), it is widely believed to be an early-onset version of aging.

Yet, few people think that nuclear membranes are the only thing that is altered in aging, as aging is generally considered too complicated for that. Instead, nuclear membranes are recognized to be one aspect within a larger pathway that is altered in aging.

Familial Alzheimer’s disease (AD) is a genetic disorder caused by mutations in APP, PSEN1, or PSEN2, which are all part of the APP processing pathway and thus (among other things) amyloid plaque production. Since it manifests in several ways that resemble sporadic AD (episodic memory loss, Aβ plaques, tau tangles), it is widely believed to be a an early-onset version of sporadic AD.

In contrast to progeria and aging, familial AD is generally thought to be a model of sporadic AD that captures almost all of the key pathways involved. As a result, one of the major justifications for clinical trials to treat sporadic AD by removing amyloid plaques is that the genetics of familial AD are all related to APP processing and thus amyloid plaque production.

There are probably several good arguments for why this progeria:aging::familial AD:sporadic AD contrast doesn’t make sense, but I still thought it might be interesting.

Making a shiny app to visualize brain cell type gene expression

Attention conservation notice: A post-mortem of a small side project that is probably not interesting to you unless you’re interested in molecular neuroscience.

This weekend I put together an R/Shiny app to visualize brain cell type gene expression patterns from 5 different public data sets. Here it is. Putting together a Shiny application turned out to be way easier than expected — I had something public within 3 hours, and most of the rest of my time on the project (for a total of ~ 10 hours?) was spent on cleaning the data on the back end to get it into a presentable format for the website.

What is the actual project? The goal is to visualize gene expression in different brain cell types. This is important because many disease-relevant genes are only expressed in one brain cell type but not others, and figuring this out can be critical to learning about the etiology of that disease.

There’s already a widely-used web app that does this for two data sets, but since this data is pretty noisy and there are subtle but important differences in the data collection processes, I figured that it’d be helpful to allow people to quickly query other data sets as well.

As an example, the gene that causes Huntington’s disease has the symbol HTT. (I say “cause” because variability in the number of repeat regions in this gene correlate almost perfectly with the risk of Huntington’s disease development and disease onset.) People usually discuss neurons when it comes to Huntington’s disease, and while this might be pathologically valid, by analyzing the data sets I’ve assembled you can see that this gene is expressed across a large number of brain cell types. This raises the question of why — and/or if — variation in its number of repeats only causes pathology in neurons.

Screen Shot 2016-06-13 at 11.35.10 AM

Here’s another link to the web app. If you get a chance to check it out, please let me know if you encounter are any problems, and please share if you find it helpful.


Aziz NA, Jurgens CK, Landwehrmeyer GB, et al, et al. Normal and mutant HTT interact to affect clinical severity and progression in Huntington disease. Neurology. 2009;73(16):1280-5.

Huang B, Wei W, Wang G, et al. Mutant huntingtin downregulates myelin regulatory factor-mediated myelin gene expression and affects mature oligodendrocytes. Neuron. 2015;85(6):1212-26.

Eight years of tracking my life statistics

Attention conservation notice: Borderline obsessive navel-gazing.

Most mornings, I start my day — after I lie in bed for a few minutes willing my eyes to open — by opening up a Google spreadsheet and filling in some data about how I spent that night and the previous day. I’ve been doing this for about eight years now and it’s awesome.

I decided to post about it now because self-tracking as a phenomenon seems to be trending down a bit. Take for example former WIRED editor Chris Anderson’s widely shared tweet:

Screen Shot 2016-05-07 at 6.07.48 PM

So this seems a good time to reflect upon the time I’ve spent self-tracking so far and whether I’m finding it useful.

But first, a Chesterton’s fence exercise: why did I start self-tracking? Although it’s hard to say for sure, here’s my current narrative:

  • When I was a senior in high school, I remember sitting in the library and wishing that I had extensive data on how I had spent my time in my life so far. That way when I died, I could at least make this data available so that people could learn from my experiences and not make the same mistakes that I did. I tried making a Word document to start doing this, but ultimately I gave up because — as was a common theme in my misspent youth — I became frustrated with myself for not having already started it and decided it was too late. (I hadn’t yet learned about future weaponry.)
  • I used to read the late Seth Roberts’ blog — it was one of my favorites for a time — and he once wrote a throwaway line about how he had access to 10 years of sleep data on himself that he could use to figure out the cause of his current sleep problems. When I read that early in college I thought to myself “I want that.”
  • In sophomore year of college my teacher and mentor Mark Cleaveland assigned me (as a part of a class I was taking) to write down my sleep and how I spent my time in various activities for a week. This was the major kick into action that I needed — after this, I started tracking my time every morning on the spreadsheet.

It takes about 66 days to develop a habit. The more complex the habit, the longer it takes. I think that by about 100-150 days in it was pretty ingrained in me that this was just something that I do every morning. After that, it didn’t take much effort. It certainly did take time though — about 3-5 minutes depending on how much detail I write. That’s the main opportunity cost.

Three of the categories I’ve tracked pretty consistently are sleep, exercise, and time spent working.

Here’s hours spent in bed (i.e., not necessarily “asleep”):

Screen Shot 2016-05-07 at 8.54.22 PM

black dots = data points from each day; red line = 20-day moving average

Somewhat shockingly, the mean number of hours I’ve spent in bed the last 8 years is 7.99 and the median is exactly 8.

Here’s exercise:

Screen Shot 2016-05-07 at 8.58.39 PM.png

I’m becoming a bit of a sloth! Hopefully I’ll be able to get this back up over the next few years. Although note that I have no exercise data for a few months in Summer ’15 because I thought that I would switch solely to Fitbit exercise data. I then got worried about vendor lock-in and started tracking manually again.

Here’s time spent working (including conventional and non-conventional work such as blogging):

Screen Shot 2016-05-07 at 8.49.54 PM

One of the other things I’ve been tracking over the past few years is my stress, on an arbitrary 1-10 scale. Here’s that data:

Screen Shot 2016-05-07 at 9.12.26 PM

In general, my PhD years have been much less stressful than my time studying for med school classes and Step 1. Although it’s not perfect, I’ve found this stress level data particularly valuable. That’s because every now and then I get stressed for some reason, and it’s nice to be able to see that my stress has peaked before and has always returned to reasonably low levels eventually. I think of this as a way to get some graphical perspective on the world.

I track a few other things, including time spent on administrative tasks (like laundry), time spent leisure reading, time spent watching movies, and time spent socializing.

I also track some things that are too raw to write about publicly. Not because I’m embarrassed to share them now, but because I’m worried that writing them in public will kill my motivation. This is definitely something to consider when it comes to self-tracking. For me, my goal has first and foremost been about self-reflection and honesty with myself. If I can eventually also share some of that with the world, then more’s the better.

Overall, I’ve found three main benefits to self-tracking:

  1. Every now and then, I’ll try to measure whether a particular lifestyle intervention is helping me or not. For example, a couple of months months ago I found that there was a good correlation between taking caffeine (+ L-theanine) pills and hours worked. Although this is subject to huge selection bias, I still found it to be an interesting effect and I think it has helped me optimize my caffeine use, which I currently cycle on and off of.
  2. There have been a few times these past 8 years when I’ve suddenly felt like I’ve done “nothing” in several months. One time this happened was about a year into my postbac doing science research at the NIH when it seemed like nothing was working, and it was pretty brutal. That time and others, it’s been valuable for me to look back and see that, even if I haven’t gotten many tangible results, I have been trying and putting in hours worked. Especially in science where so many experiments fail, it’s helpful for me to be able to measure progress in terms of ideas tried rather than papers published or some other metric that is much less in my control. GitHub commits could also work in this capacity for programmers, although that’s slightly less general.
  3. The main benefit, though, has not been my ability to review the data, but rather as a system for incentivizing me to build process-based habits that will help me achieve my goals. I enjoy the bursts of dopamine I get when I’m able to write that I worked hard or exercised the previous day — or that I got a lot of high-quality socializing in with friends or family — and it makes me want to do that again in the future.

Do you want to try a similar thing? Check out this blank Google spreadsheet for a quick start; it has a bunch of possible categories and a few example days for you to delete when you copy it over to your own private sheet. I like Google sheets because they are free and able to be accessed anywhere with an internet connection, but it’s certainly not a requirement.

Even if you don’t try it, thanks for reading this essay and I hope you got something out of it.


Arterial aging and its relation to Alzheimer’s dementia

I’m a big proponent of the role of arterial aging in explaining dementia risk variance, in large part because it explains the large role that vascular-related risk factors have in promoting the likelihood of Alzheimer’s disease (AD). However, some data suggests that the burden of ischemic events and stroke cannot explain all of the vascular-related AD risk. Recently, Gutierrez et al. published a nice paper which suggests that non-atherosclerotic artery changes with age may explain some of this residual vascular-related risk of AD. In particular, they used 194 autopsied brains and found five arterial features which strongly correlated with aging, including decreased elastin and concentric intimal thickening. Importantly, these features also correlated with AD risk independently of age.

The authors propose that the arterial aging features are a consequence of mechanical blood flow damage that accumulates over the years. If it is true that the damage is mechanical, it suggests that it may be difficult to reverse with existing cellular and molecular anti-aging therapies. For those people who are interested in slowing down aging, the brain must be a top priority because it cannot be replaced even by highly advanced tissue engineering approaches to replace the other organs. Thus, this sort arterial damage needs to be addressed, but to the best of my knowledge it has not been, which is one of the many reasons that I expect that serious anti-aging therapies are much further out than are commonly speculated in the popular press.

Are four postulated disease spectra due to evolutionary trade-offs?

I recently read Crespi et al.’s interesting paper on this subject. They describe eight diseases as due to four underlying diametric sets that can be explained by evolutionary/genetic trade-offs:

  1. Autism spectrum vs psychotic-affective conditions
  2. Osteoarthritis vs osteoporosis
  3. Cancer vs neurodegenerative disorders
  4. Autoimmunity vs infectious disease

Of these, #2 and #4 seem obviously correct to me based on my fairly limited med school exposure, and they describe the evidence in a systematic way. I don’t know enough about the subject matter to speculate on #1, but I would like to see more genetic evidence.

Finally, I found their postulated explanations for #3 somewhat weak and I personally think that it is a selection bias trade-off, i.e. a case of Berkson’s bias as applied to trade-off. That is, since both cancer and neurodegeneration are age-related conditions, you could think of aging as the “agent” that selects either neurodegeneration or cancer as the ultimate cause of age-related death. I could be persuaded to change my mind on the basis of genetic predisposition evidence or some other mechanism, but I found the mechanism of apoptosis to be weak since apoptosis occurs (or doesn’t occur when it should) in many, many diseases, and moreover it is far from clear that neurodegeneration is mostly due to apoptosis as opposed to some other mechanism of cell death. A mechanism that might be most persuasive to me is one related to immune cells, since they clearly play a large role in regulating cancer growth, and also have high expression for the most GWAS risk factors for Alzheimer’s disease. But I still suspect that the selection bias is primary.

How to do Bayesian shrinkage estimation on proportion data in Stan

Attention conservation notice: In which I once again document my slow march towards Bayesian fundamentalism. Not of interest unless you are interested in shrinkage estimation and Stan.

As I describe in my essay on trying to determine the best director on imdb, you usually can’t trust average ratings from directors with a small number of movies to be a good estimate of the “actual” quality of that director.

Instead, a good strategy here is to shift the average rating from each director back to the overall median, but to shift it less the more movies that person has directed. This is known as shrinkage estimation, and in my opinion it’s one of the most underused statistical techniques (relative to how useful it is).

The past few weeks I’ve been trying to learn the Bayesian modeling language Stan, and I came across a pretty good model for shrinkage estimation using a beta-binomial approach in this language (described in 1, 2). Here’s the model, which uses batting averages from baseball players.

In order to determine the amount of shrinkage in this model, I plotted the “actual” (or “raw”) average versus the estimated average using this model, and colored the data points by the log of the at bats (lighter blue = more at bats).

Screen Shot 2016-03-10 at 7.39.03 PM

As you can see, players with more at bats have less shrinkage. At the extreme, two players who are 0/1 on the season still have an estimated average of ~ 0.26 (which is the median of the “actual” batting averages).

Notably, there are fewer players whose averages are decreased due to the shrinkage estimation than the reverse. Perhaps managers are inclined to give players a few more shots at it until they prove that their early success was just a fluke.

A clinical trial for omental transposition in early stage AD

A couple of years ago I wrote about treating AD with omental transposition, a radical therapy with success in ~ 35% of patients in one case series. Today I just noticed that there is a non-randomized, single-arm clinical trial on its use in patients with early stage AD (MoCA score 11-18), in Salt Lake City, UT. Estimated study completion date: May 2019.

This is especially interesting because they have a relatively thorough explanation of how the surgery works. In the general surgery portion of the procedure, an omental flap is created, which receives blood supply from the right gastric and gastroepiploic arteries. Next, a subcutaneous tunnel is created that travels up the chest wall and neck to behind the ear.

In the neurosurgery portion of the procedure, a portion of bone is removed near the temporal-frontal area, followed by removal of the dura and arachnoid membrane. The omentum is then placed on the parietal-temporal-frontal area of one cerebral hemisphere, and connected to the dura via a suture.

Besides this tissue grafting approach, other neurosurgical approaches to Alzheimer’s have included:

  1. CSF shunts (to the atria or ventricles)
  2. Intraventricular infusions (of bethanecol, NGF, or GM1)
  3. Gene therapy with infusion of NGF-expressing cells
  4. Electrical stimulation (of the vagus nerve, nucleus basalis of Meynert, or the fornix)