Being an anal scientist

I think there's a strong correlation between being excessively anal and getting good results in lab.  Certainly, the vast majority of successful lab scientists I've met are just super fussy about their experiments.  In fact, I know one guy who used to give his undergrad very precise instructions on exactly how to vortex an eppendorf tube of liquid.  It's so annoying!  And yet sweating these seemingly insignificant details somehow seems to correlate with getting good results, especially in biology, where stuff doesn't work if you look at it funny.

Of course, 90% of the stuff that fussy people fuss about probably doesn't matter at all.  But that reminds me of this quote attributed to advertising pioneer John Wanamaker: "Half the money I spend on advertising is wasted; the trouble is I don't know which half."  The thing with experiments and fussy people is that you know that 90% of the fuss is completely irrelevant, but you never know which 90%...

Planets and cells

Spend enough time looking at cells under a microscope and they’ll start to look like little planets, teeming with a hidden life of their own.

Pretty much the same thing, right? Looking at Earth, the picture is familiar. The shape of the continents, the cities marked by the light- all of these things and more are immediately understandable because of our experience. We know that the light represents people living in clusters, defined by political, economic and geographic constraints.

The cells, on the other hand, are more enigmatic. They are immortalized early B cells from the GM12878 line, one of the most well studied and deeply sequenced cell lines to date. Yet, in comparison, the mRNA spots are mysterious little lights whose meaning we are only beginning to decipher.

The scale of a system is a curious thing. When we think about our planet and how it got to that state of light that is so familiar, the answer seems simple enough. Indeed, historians have documented in extensive detail the various ways in which interactions between civilizations, empires, countries, and corporations ultimately led to the present state.

But what if we didn’t know any of this? What if we were gargantuan beings and saw the earth through special microscopes like we see cells? What sort of narratives might we invent to explain the pattern of light? Would we capture any of the nuances that are so obvious to us now?

For me, it's a fun thought experiment to think about what I might conclude from my "earth cell". I'd likely define the light as an autonomous entity- completely oblivious to the human activity that drives it simply because I could not measure it. I might hypothesize some repelling factor for the dark blue areas since light never appears there without ever understanding the difference between land and sea. Though I wouldn't have the slightest notion of what a port is, I might conclude that the light is enriched along the borders between the light blue and dark blue areas. I could make these observations, and though correct, they completely undercut the depth of understanding that we take for granted.

So then, going back to the cells, it's mind-blowing to imagine what is "really" going on. Systems are crazy things. Sometimes, it seems like they are far more than the sum of their parts, leading to incredible complexity from a few simple rules. Other times, it seems like they are much less, filtering a wide range of inputs into a few reliable outputs. And somehow, they repeatedly emerge in nature no matter what scale you look at. For now, I find those little spots deeply intriguing, and I think the only certain truth is that too much time in a dark microscope room can make yourself a little loopy...

A bit of info on RNA-seq

So we've been playing around with comparisons between RNA-seq and RNA FISH, and while we're still a ways away from a more comprehensive study, I thought I'd share a little tidbit.  There are two sorts of comparisons one might want to make when quantifying gene expression via RNA-seq.  One is comparing expression of a single gene in different conditions; i.e., find the "hits".  We have found this to be reasonably good at finding hits, probably with a validation rate of around 50% or more.  The magnitude of the hits are, well, sort of close to what the RNA-seq would predict.

The other is to compare the expression of different genes in the same sample.  Here, the relevant question is whether genes with similar read counts (FPKM/RPKM) have the same actual number of RNA as measured by RNA FISH.  Basically, what's the correlation between RNA-seq and RNA FISH?  We've found that the correlation is quite good–but with an important caveat.  The correlation is good for high expressing genes.  This means that if you have twice as many beta-actin FPKM as GAPDH FPKM, chances are you'll have roughly twice (ish) the actual number of beta-actin as GAPDH mRNA in your cells.  However, as you get down to below around 50-100 FPKM, there is much less correlation and way more noise.  That means that if you have Transcription Factor A and Transcription Factor B with FPKMs of 25 and 50 respectively, pretty much all bets are off as to whether TF A or TF B will have more actual mRNA numbers.  Usually, people seem to care more about the expression of these low genes anyway, so it's important to keep this in mind.  Another point is that if you're getting low FPKM values (I'm going to be vague about how low), there's a good chance that the RNA is not actually expressed at all in the cell line you're looking at.  We've seen it happen a lot.  A LOT.  You know who you are... :)

Anyway, hope this information is useful to someone, somewhere.  I promise we'll have a lot more on this sometime soon.

On 23andMe

We've been thinking about SNPs in the lab lately, and it made me sort of curious about my own genotype. I'd been contemplating ordering a kit from 23andMe, a company which analyzes your spit-derived DNA on a SNP array. The recent price-drop to $99 made the decision pretty easy. So, now that I have my results, were the $99 well spent?

Let me start by saying that, despite 23andMe's marketing ("take a more active role in managing your health" ), I was under no illusion that I would learn anything too useful about my health. After all, most of the SNPs associated with (non-Mendelian) traits explain very little of the actual variation in the population. For example, the "Height" SNPs on 23andMe only account for, like, 0.6 inches of human height variation. It was cool, though, to learn that I'm not a carrier for any of the 50 or so Mendelian diseases (ex: Cystic Fibrosis, Phenylketonuria) listed on the website.

The most surprising thing I saw was that, according to my genes, I should almost certainly be lactose intolerant---which I know I’m not. The website has a neat "community" feature for each SNP, where people can discuss whether they actually have the associated phenotype (the page for the "intelligence" SNP is pretty funny…with lots of people defensively posting their IQ scores).  On the page for lactose intolerance, I saw tons of people post something along the lines of “Yup, but I was fine until my late 20s!” So, at the very least, I’ve learned that I should binge on ice cream and cheese while I still can...

Part of the problem with interpreting the “usefulness” of health results from 23andMe is that there’s so much random stuff that there’s sort of a dilution effect for what I can get worried about. For example, if I had only learned that I have a 34% increased risk of developing prostate cancer (the average probability is 17.8%, and my SNPs supposedly push me up to a 23.9% risk), I might be a bit shaken up. But I’m also twice as likely than average to get exfoliation glaucoma and 35% more likely to get atrial fibrillation. And then I’ve got to balance my worry with the relief that I’m 25% less likely to get venous thromboembolism, and I have a 12% decreased risk of getting gallstones. I calculated that, umm, I should be approximately 0.43522% more worried about my health now than I was before I saw my results. 

I guess I was most excited to learn about my ancestry. The website maps certain SNPs to geographic locations 500 years ago, when rapid migration did not yet exist (nice explanation of how they estimate ancestry). I knew, generally, that my ancestors were from Gujarat, India---a central, coastal state separated from the Arabian peninsula and the Horn of Africa by a few hundred miles of sea. This area was highly accessible by the European colonialist explorers of the mid-20th century, and it's not too far south of the Silk Road, the ancient trading route that linked the Far East with the Middle East. The point of the contrived geography lesson is that it convinced me that I had to have recent ancestors from beyond the Indian subcontinent. In addition, I recalled that my grandfather thought he had an English grandfather, though it was a murky claim because the alleged Brit died very young, and none of our living relatives could corroborate it confidently. Anyway, this was all was thoroughly debunked by my SNP data.

So I'm more Indian than a plate of Chicken Vindaloo (which is actually sort of Portuguese).

Whether or not "personalized medicine" takes off in the near future, there's a thrill to being an early adopter, so I certainly don't regret my purchase. But I'm not sure I'd recommend it to people who are on the fence, because, chances are you won't really learn all that much.

A really nice publishing experience

These days, it really feels like the publishing system is in pretty bad shape, especially peer review.  I still wonder whether the system makes sense, but I did want to report that we had a really nice time at Genes and Development.  The reviews were decent, everything went super fast, and the editors were decisive and really managed the process well.  Highly recommended!

Gautham and Priscilla's wedding tea party

Congratulations to the happily married couple!

It was a beautiful day at near Belmont Mansion out in the Philly burbs, filled with food, music and dancing...

From left to right: Arjun (with Anaya), Marshall, Patrick, Gautham, Priscilla, Justin Bieber, Justina Bieber, Maggie, Paul, Chris.