Cilantronomics

We had dinner last night in one of favorite local eateries, a wonderful little Mexican place in a neighboring town. When I sat down, my eye was drawn immediately to my sort of dish -- one with a rich sauce combining the tang of tomatillos with the zing of cilantro. I picked well.

I really do love cilantro. Despite an extensive garden growing up, it was only in my adult life that I encountered this herb. I've been making up ever since. It works well in so many situations, not only in Mexican but also a lot of Asian cooking. The excellent Tibetan buffet in Central Square uses it extensively, particularly in a salad that works equally well before the meal as after, with the bite of cilantro contrasting with sweet cherry tomatoes and mango chunks. I even have a pot of it on my desk, which I share with my neighboring cilantrophiles.

However, not everyone loves cilantro. And it isn't just some folks might not like that little edge -- no, for some it tastes awful. Rather than some herbal bite, they taste soap. Or weirder. What other herb has its own http://www.ihatecilantro.com/?

Is it genetic? Alas, there has been a dearth of research on cilantro tasting -- indeed, it doesn't seem to rate an OMIM entry. There is a compound called PTC which is known to untastable by some, including this correspondent, and is genetically linked (my father can taste it; haven't surveyed the rest of the clan). With the help of some research by one of my office neighbors (and fellow cilantro fan), I did learn that 23andMe includes cilantro taste in their questionnaire. It isn't clear whether the other public and private genome projects are tracking this key phenotype.

Okay, I jest a bit. But while the ability to taste cilantro, or PTC, or the host of other innocuous traits which are staples of grade school genetics labs (e.g. widow's peak, hitchhiker's thumb, attached earlobes, etc) aren't exactly critical to understand, they will be interesting to understand. Widow's peak doesn't change someone's life, but to understand it is to understand a bit more about how patterns are laid out. The sciences of smell and taste have advanced tremendously over my lifetime; a whole new taste was found! Identification of smell receptors (recognized by a Nobel) and taste receptors have given great insights -- but we still understand very little.

Are there practical applications for smell & taste research? Of course. But to me the most interesting part is to figure out how it works. PTC doesn't seem so complicated, as the test paper doesn't have any flavor other than paper. But cilantro seems like a much more complicated, and interesting, question. Why does it taste bad rather than just not taste?

Is there an underlying soapiness which I just don't taste? In this case, tasters have a receptor for the magic compound (which is what?) and non-tasters simply lack it. Or does a different receptor bind the compound in tasters, in which case they have a gain-of-function mutation? Or, perhaps they have a partial loss of function -- there are a number of known compounds with concentration-dependent odor, probably due to differential binding to different receptors. In other words, at low concentrations these compounds bind to high-affinity receptors (yielding one perception) and at high concentrations some additional one Or, perhaps a partial gain of function in the non-tasters -- the same model could apply.

No, I wouldn't recommend basing an R01 application on the science of cilantro taste. Nor is it likely to tease a few million from some VCs as the core of a business plan. Cilantro haters will probably never have the option of genetic therapy to alter their perception. But it is still an interesting scientific question, and I look forward to personal genomics shedding some light on it.

The Blue Bus Grows Up

A striking characteristic of the Cambridge biotech scene is how it is concentrated in an urban setting. While there are a lot of biotechs elsewhere in Massachusetts, the Hub's hub is clearly a 2+ mile long zone. One challenge this offers is getting to work via Boston's transportation system.

Boston doesn't have an awful transportation network, but it isn't golden either. The transit system is decent, but the routes still largely follow a radial design, with routes that have changed little in the last half century (I kid not; I've seen a map that old & it takes a careful eye to find the differences). An extensive network of commuter rail feeds the downtown, but is split between two termini separated by a mile. The highway network has a number of gaping gaps, due to a mass cancellation of uncompleted highways in the early 1970's. However, this wasn't necessarily bad for biotech; I've spent half my career in offices that would literally be in the middle of the road should those highways have been built (for example, this very different vision for 640 Memorial Drive than a genomics-based pharmaceutical company).

The commuter rail option presents a particular challenge. One station, South Station, is connected to the Red Line subway which has two stops (Kendall & Central) proximal to many biotechs. The other station, North Station, has terrible connections to Cambridge, other than the perhaps future expansion of the zone into the Cambridge-Charlestown-Somerville interzone. But, if you live north of town it's either deal with getting from North Station to Cambridge or brave I-93. So, by multiple subway connections or a tortuous pedestrian path through Mass General's campus to the Red Line, a not tiny cohort of biotechies has made their commute this way.

Then about five years ago a new option appeared. Little blue buses promising a single seat ride from North Station to Cambridge, with a twisting route designed to be near nearly every major employer in the zone. Called EZRide, for $1 anyone can ride, but better yet the larger employers offer ride-all-year stickers.

The service was a bit slow starting up & went through a few hiccups, but over time it has been impressive. If memory serves, the initial frequency was every 30 minutes; this has been steadily dropped so that now a bus shows up every 8 minutes during commute time. However, demand has grown even faster; during core commuting times the bus is at its legal limit, with only ~30 seats and less than a dozen legal standees.

But a couple of weeks ago the announcement came out: a new vendor would be running full size buses on the route. And last week they showed up. On the one hand, there are more seats -- but not as many as one might think due to the layout. On another, more legal standees and less of the EZRide shuffle -- having to exit the bus at the early stops to let people off, as if you were standing you were a cork in aisle of the old buses. The longer buses don't handle the tight turns as well but do away with the most unpleasant aspect of the old buses: their short wheelbase combined with Cambridge's potholes yielded an amusement-park quality bumpy ride (particularly unpleasant if you made the mistake of leaning against the wheelchair lift).

EZRide isn't run by the transit system, but rather by a quasi-public entity called CRTMA which is charged with improving transit into Cambridge. The director, Jim Gascoigne, is energetic and personable and often on the scene, particularly when weather or accidents snarl require emergency re-routing.

In some ways the EZRide highlights issues in Boston. The T does an okay job, but it apparently never occurred to them in several decades that a market existed for a route from North Station to Cambridge. I've also seen the truly surreal quality of Boston from the blue bus: due to some construction, one Boston police officer directed the bus to stop in a new location, where the driver was promptly berated & ticketed by a second Boston officer. This is the town where the mayor had major apoplexy when a nearby airport added Boston to their name; transportation issues are about turf battles as much as moving people around. The planners also have a fondness for expensive megaprojects (the current shopping list can be found here). Several of these would have important benefits for the biotech zone -- for example, the proposed Urban Ring would run right through it & connect the zone to the Longwood Medical Area.

However, perhaps what is more realistic are more EZRide-like services, perhaps connecting to the south (Brookline, Brighton/Allston) that have surprisingly poor connections, or to the large transit hubs to the north (Wellington, Anderson). A direct connection to Charlestown wouldn't be a bad concept either.

In the meantime, I'll keep riding EZRide. And anxiously awaiting my train line(s) getting the free WiFi service a few lucky commuters have gotten to pilot. One more good reason to stay off the road and out of my car!

Panda genome arrives

China announced over the weekend the completion of the giant panda genome.

For the benefit of presidential candidates who can't conceive of the value of scientific research on bears I'll suggest a few questions worth exploring in the panda genome (beyond the obvious direction of weapons development)

First, the panda genome is one more mammalian genome to add to the zoo. For comparative purposes you can never have too many. Since other carnivore genomes are done (first & foremost the dog, but cat as well), this is an important step towards understanding genome evolution within this important group. It is the first bear genome, but with the price of sequencing falling it is likely that the other bears will not be in the extremely distant future (with the possible exception of Ursa theodoris).

Second, completion of a genome gives a rich resource of potential genetic variants. In the case of an endangered wildlife species such as panda, these will be useful for developing denser genetic maps which can be used to better understand the wild population structure and the gene flow within that structure. Again, if you are running for president please read this carefully: this has nothing to do with paternity suits. If you want to manage wildlife intelligently and make intelligent decisions about the state of a species, you want to know this information.

Third, pandas have many quirks. That bambooitarian diet for starters. Since they once were carnivores, it is likely that their digestive systems haven't fully adapted to the bamboo lifestyle. Comparisons with other carnivores and with herbivores may reveal digestive tract genes at various steps in the route from meat-eater to plant-eater.

Fourth, as the press release points out, there are many questions critical to preserving the species which (with a lot of luck) the genome sequence may give clues to. First among these: why is panda fertility so low? U.S. zoos have been doing amazingly well in this century, but that's only 4 breeding pairs. The Chinese zoos have many more pandas & many more babies, but it's going to take a lot more to save the species.

Great Galloping Gerbils!

An item on CNN mentioned that satellite technology will be employed to monitor the endangered California Kangaroo Rat. This reminded me of a Nature paper this summer I meant to mention, because the image blew me away (plus it's the first time I've seen Google Earth used as a source for a scientific paper!).

The paper is about models of disease spread (these gerbils are reservoirs for plague), but the thing which jumped out was the Google Earth image; the two images above are from around the same region of Kazakhstan. The gerbils clear vegetation from around their burrows, and these burrows are in huge complexes. The more zoomed in image above is several kilometers wide and yet is packed with gerbil burrows. If you have Google Earth and look around 44.766991 76.449699 you can zoom way out and still see the gerbil complexes. I saw some huge prairie dog towns out west when I was a boy, but nothing on this scale!

How many animals leave traces which can be seen from 30+Km up (the image quality is uneven for this region of the world in Google Earth -- clearly shots are merged from different seasons and resolutions, but 30Km is a conservative estimate)? Human activity obviously. When I think of animal-built structures I generally jump to beaver dams or termite mounds, but they aren't nearly this extensive.

The Joys of DIY Dynamic Programming

For nearly the first decade of my bioinformatics career I carried around a dirty little secret -- well, at least at times I felt it was one. I had coded many things, I could explain many algorithms, but I had never coded a dynamic programming alignment algorithm -- the core to so much I did. I had slightly hacked one version (just to have it do an all-all comparison of a database, doing each possible pairing only once). Finally, for a bunch of reasons, I sat down and did it -- my very own Smith-Waterman implementation.

I'm reminded of this because a couple of weeks ago I rolled back my sleeves and knocked one out again. Now, just the fact I did this reveals a bit about me. I did find at least two freely available C# implementations on-line (e.g. the C# version of JAlign) and there is a plethora of C implementations. There is also Ewan Birney's magnificent Dynamite, pretty much the catch-all for the field (Dynamite is a programming kit for doing this; in effect a programming language for dynamic programming). But, partly as a point of pride & partly because I saw I'd need to hack the one C# copy I looked at in detail, I did it. I even wrote a schmancy version -- a simple cDNA to genomic sequence aligner with two classes of gaps (one being an intron, with a really trivial model of a splice junction -- I think it used dinucleotides) All coded in Perl -- no speed demon, but it solved the problem where we needed it.

Now, it took me a good few hours to do it -- better than the few days of the first time, but not instant. I can claim that this time I didn't fall back on any study aids, such as the many online descriptions or Eddy & Durbin's & co. very well written book.

The implementation says a lot about me too. I thought of many ways to code it and finally settled on one. For example, there is the question of how to represent the alignment matrix; I used a two-dimensional array scheme (actually implemented using dictionaries -- a holdover from my Perl-centric days) but I could have also made it a graph of nodes. There is also the actual thrashing through the matrix -- the algorithm is inherently recursive, but following familial idiosyncracies I wrote the code to use loops -- well, actually I completely waffled and implemented so it can use recursion, but actually loops through! The applications I'm considering are going to be short alignments, so I didn't worry about memory efficiency (who wants to be that will bite me back!) nor did I fixate on speed (care to double the bet?) -- indeed, I wrote it to allow all sorts of baroque variations, such as different penalties for opening gaps in the two different sequences & for basic profile-to-sequence alignments. Plus it is either Smith-Waterman (local) or Needleman-Wunsch-Sellers (global), with a simple toggle.

So now the pitch: If you are a bioinformatics programmer & you haven't written one, I urge you to do it. It's great practice & nothing illustrates an algorithm like trying to implement it. If you don't consider yourself a programmer, guess what? It's perhaps not the obviously easy first start, but just thinking about it will stretch your mind. Plus, you get a free bioinformatics Rorschach test from your implementation choices!

One last thought: who can think up (and execute) the most comically baroque -- but functional -- implementation of S-W/NWS? Has it already been done in PostScript? How about in a relational database (I've written some pretty baroque SQL this year, but I doubt I could tackle this)? S-W as an Excel spreadsheet? Coded with glider guns? A full description for a true Turing machine? Of course, the grand prize winner would clearly either be to build a DNA computer to compute an alignment -- but perhaps that could even be topped by implementing the algorithm with living cells as the alignment cells!

One more Olympic thought

One other item that was in the mental draft of yesterday's Olympic pondering, but was inadvertantly dropped. Another possible genetically-driven edge in athletic performance would not be directly on performance but on the reaction to performance. Prime athletes might have different pain or endorphin responses, less post-exercise inflammation, different injury responses. Some of these might be specific to specific events or types of sports -- joint pounding running or gymnastics puts very different stresses on the body than something like swimming or speedskating.

An Olympic Pondering Decathlon

Okay, my biannual stint of Olympic watching is about to conclude. A bunch of speculations suggested by this year's stretch, starting with the utterly unscientific and ending with more genomic oriented queries.

1) Having now watched two Olympics using a Digital Video Recorder, it's completely clear that having a few fast forward speeds is no way to navigate multi-hour recordings to find what you want. Surely there are better UIs for this! The thumbwheel on an iPod is one obvious choice, but there must be other ways.

2) The Summer games are blessed with multiple events which touch on multiple disciplines: the decathlon, heptathlon, modern pentathlon & triathlon. Why isn't there a true multi-discipline Winter sport? Biathlon is a glorious combination of two diametrically opposed skills -- racing and precision shooting -- and there is also the Nordic combined, but neither of these sample a wide range. How about this for a Winter hexathlon:
A) 500m long-track speedskating
B) 2500m long-track speedskating
C) Downhill skiing
D) Slalom
E) 10Km X-C skiing
F) ski jump (small hill)

3) I once contemplated attending a HUPO meeting in Beijing; atop an interesting program there was a post-conference trip option to tour the country. There were two issues: I'd have to foot my own travel expenses & I'd be in serious hot water at home for visiting the Wolong panda center solo. But, now I'd definitely go -- particularly if they replaced a typical dry scientific kickoff with another Zhang Yimou spectacular, I wouldn't hesitate!

4) As a kid I did occasionally have Olympic daydreams. I'm a bit over the hill now, but between athletes older than I such as Dara Torres and hearing that some countries will field just about anyone, perhaps I gave up too soon. If I could pick anything, it would be long track speedskating, the most graceful speed sport bar none. But, more realistically perhaps I could go for the 1500 meter freestyle swimming -- I'd estimate my time is off by only a factor of 3 -- perhaps with some regular training I can get that down to 2!

5) Of course, even with some extensive training I'd look pretty odd on the blocks -- I'm 5'8" and from what I can tell in the TV coverage, it's a rare swimmer who isn't a few inches over 6'. Clearly there are advantages to height in a large number of sports -- but I'm also clearly too tall for a shot at women's gymnastics (atop the other obvious issue). It would be interesting to see which sports have the highest and lowest dispersion in athlete height -- and what those patterns look like. What sport should I have chosen based only on my height?

6) During the Olympics, the world's tallest living woman died, after a life with many health difficulties. Clearly, there are limits to the advantages to height. What sport has the tallest athletes?

7) What more subtle anatomic characteristics might lead to athletic advantage? Differences in muscle fiber composition are an oft-cited one. A TV profile of superswimmer Michael Phelps claimed he is 'double jointed'. But what else. For example, are there subtle differences in some individual's lungs which lead to more efficient air exchange? Smoother surfaces on bone joints?

8) Diving lower, are there biochemical differences? Again, could there be differences in oxygen transfer or usage? Differences in energy metabolism?

9) If we did genome screens of the athletes, what SNPs would we find over-represented? How many of those would be 'obvious' and how many would lead to new genes which influence performance? Already there is at least one company offering genome scans to predict what sport you should stuff (er, steer) your kid into.

10) The sad story of Flo Hyman illustrates another aspect of selection for unusual body types: she died of a aortic dissection due to Marfan's syndrome, which probably also led to her tall, thin stature which was an advantage on the volleyball court. What other genetic variants have a dicey risk/reward trade-off in the athletic arena? And how many of these are a serious medical issue for regular folks?