Saturday, August 31, 2013

GO analysis by "greedy" assignment of genes to categories

- Gautham

This post proposes a simple method to analyze gene hits from a screen in terms of GO categories, provides an example from the literature, and links to R code.

It is easy to find many very sophisticated algorithms and frameworks for analyzing the results of high throughput screens in the context of biological annotations to categories like the Gene Ontology (GO). The result is almost always p-values and enrichment scores for all GO categories, and it is common practice to then report the "most significant" (lowest p-value) terms. There are often a very large number of terms that meet reasonable significance criteria even after multiple hypothesis testing correction. This may be partly due to improvement in experimental methods, so that now even small effects result in statistically significant gene hits. Many comparative RNA-Seq experiments yield thousands of differential expression gene hits. 

It is tough to get a feeling for the make-up of the hit gene list by a vanilla GO analysis because of the following problems:
  • Screening by p-value tends to give large categories. This is for the same reason that screening differential expression by p-value tends to give genes with high average expression.
  • Screening by enrichment gives very small categories. Again, for the same reason as screening differential expression by fold change tends to give genes with very low average expression.
  • GO terms are not mutually exclusive, and hit lists tend to be made up of overlapping GO categories whose high rank on the list is due to a common set of hits.
The first two problems are common to any differential expression screen, but the overlap problem is unique to category analysis. Without exaggeration, these are the top 10 hits by p-value for a set of gene hits from a screen we did in the lab:
  1. multicellular organismal process
  2. anatomical structure morphogenesis
  3. system development
  4. multicellular organismal development
  5. anatomical structure development
  6. developmental process
  7. organ morphogenesis
  8. signaling
  9. skeletal system development
  10. nervous system development

More sophisticated GO analyses take into account the graph structure of GO to try to get around this problem, but it seemed like something very simple could do the trick instead. Here is a proposal:
  1. Decide the scale of GO terms you want to consider at the outset. Only consider GO terms that have more than some minimum number of genes annotated to it. The larger the cutoff, the broader the picture you will get. 
  2. Find the GO term with the highest concentration of hits. This category has the greatest number of hits in the screen per annotated gene.    
  3. Assign these hit genes to that GO category. Remove these genes from further consideration.
  4. Repeat 2,3,4 but considering only hits that remain unassigned. Continue until all hits are assigned, or you have as many GO terms as you want. Usually, the parent biological_process category will eventually get selected and sweep up all remaining genes.  

Step 1 forces you to explicitly choose your resolution. Step 2 seeks to find categories that are as likely as possible to be actually involved in the biology you are studying. Step 3 bravely (or foolishly) assigns genes as doing the most enriched function they are annotated for, and Step 4 eliminates the GO overlap problem. This kind of algorithm is commonly known as "greedy," so we refer to it sometimes as greedy GO.  

To illustrate we used the data set generated by Marks et al. 2012 (PMID: 22541430). These authors looked at the transcriptome differences between mouse ES cells maintained in serum-containing media or media containing two small molecule signaling inhibitors. They found nearly 1500 genes significantly more highly expressed in 2i media than in serum, and present their own category analysis (PANTHER database) in their Fig. 1c. Using the BioConductor-maintained org.Mm.eg.db annotation database for mouse as a source for GO annotations, here is a breakdown by the greedy method of the "high in 2i" gene hits from Marks et al. Cell 2012:


GO breakdown of 2i hits, size cutoff = 300
Num.hits.assigned Num.hits Num.genes
cellular lipid metabolic process 73 73 495
regulation of immune system process 49 54 393
cellular nitrogen compound biosynthetic process 40 44 328
ion transmembrane transport 41 45 374
carbohydrate metabolic process 43 58 414
neuron projection development 33 38 382
biological_process 1004 1283 12243

These are not the top hits, or handpicked in any way. They are the complete output of our greedy algorithm. No p-value threshold has been set here. In fact, the method does not even compute p-values. Compare to Fig. 1c in the paper, and you'll see the categories mostly agree, and that we avoided duplicate entries for lipid metabolism. The approach can be augmented with p-values by running something like goseq, which calculates the p-values for every GO category.


The R code for the functions that do the computations is here:
https://gist.github.com/gauthamnair/6400111

A script that does an analysis of the Marks et al. data is below, showing a simple way to get GO annotations into the right format, by exploiting a function in the goseq package.
https://gist.github.com/gauthamnair/6400293



Wednesday, August 28, 2013

The royal (scientific) jelly

I only recently heard about how honeybees make this stuff called royal jelly.  It's amazing!  Basically, when the queen dies, you give some other larvae large amounts of royal jelly, which leads to those larvae turning into queens (epigenetic effects, apparently).  Super cool.

I also just read Malcolm Gladwell's "Outliers", and the basic idea is that people who end up being great successes owe as much to circumstance (i.e., "royal jelly") as to inherent talent and ability.  I largely agree with him.  And the cool thing is that he has at least some hard evidence to back up this theory.  One example is that most pro hockey players tend to be born early in the year.  The idea is that (due to age cutoffs in the leagues), the January born kids end up playing with the December born kids of the same calendar year, and since they're bigger and stronger, they end up getting the attention of the coaches, etc.  Total royal jelly effect.

It got me thinking about royal jelly effects in science as well, for which I think there is evidence on multiple levels.  Consider the graduate students of a particular PI who went on to have successful academic careers.  Anecdotally, these often seem to be among the first students that PI has.  The natural explanation is that these students get the PIs royal jelly: the PI is heavily invested in them and has a lot of time to spend working with them.  This is absolutely not to say that these students are not very talented in their own right, nor that PI jelly alone would make a superstar out of a fundamentally mediocre student, nor that an established PI can't churn out exceptional students.  But I think it just biases things towards success.

Same goes the other way.  If you're a PI who ends up at an institution with a name that attracts talented graduate students, you will of course do better.  A great student will naturally execute a project better than a so-so one.  But more than that, I feel like good students can take ownership of their project, freeing up tons more of your time as a junior PI because you don't have to sweat all the details.  And yes, I feel extremely fortunate to have an exceptional team of people in the lab that I consider primarily responsible for our success.

Circumstance can also affects one's scientific career in other, broader ways as well.  For instance, I seem to have started my lab in a period of extreme competition for research funding.  How will this impact the trajectory of our lab?  I'm not sure.  Another thing is that systems biology is now a relatively mature field.  Long gone are the times when tagging some protein with GFP and judicious use of error bars would get you a high profile paper in Nature noted for "a novel combination of experiment and theory" (I'm kidding, kidding!  Sort of.).  Not that it's a bad field to get into by any means; I would certainly argue that it is still very vibrant and exciting, and in fact, I think that now is the period when systems biology is best poised to have a real impact on biology.  But it's just that the wind is not filling your sails out quite as much as it was 10 years ago.

I think you can also see evidence for the power of circumstance in scientific fields and nationalities.  For instance, there are tons of very talented Israelis in systems biology.  I've heard people ascribe this to all sorts of things, including the fact that Israelis have to go through compulsory military service (can't remember the logic behind that one).  But consider the following.  I also noticed a while ago a blip of very talented Turkish biophysicists.  Now is it something about the Turkish national culture that spawns biophysics?  Dunno.  But I did ask a Turkish biophysics person I knew about it, and I think he said something like "Well, all the good students go through the same institute in Turkey, and we looked at who was successful, who to look up to.  And there was Ahmet Yildiz, who was in biophysics."  Again, purely anecdotal, but could the Turkish biophysics craze all be a result of Ahmet Yildiz's work in single molecule biophysics (awesome and highly influential in its own right, by the way)?  Now that is some serious impact!  Maybe the reasons that there are so many Israelis in systems biology has its roots in Uri Alon and Naama Barkai's early pioneering work in the field?  Could be.  Life truly is random.  I got into the field because of a random chat with Sam Isaacson, who was also a graduate student with Charlie Peskin at the time, that got me interested in stochastic gene expression.  I worked in a lab during my math PhD because Charlie Peskin happened to play tennis sometimes with Fred Kramer, who co-ran a lab with Sanjay Tyagi (who became my other PhD advisor, and gave me plenty of royal jelly, by the way).  My paper with Hedia and Dave Dubnau happened because I had to take a train home early for physical therapy and saw Dave reading some papers by Michael Elowitz.

(Incidentally, things after that seem on the face of it to be a bit more deterministic.  But even looking at that a bit more closely, there's a lot more randomness than I thought.)

Again, none of this is to say that the people who are successful scientists are not highly talented people–they are, in an absolute sense.  It's just that there are many talented people, and who knows when the right circumstances come together to have that talent develop down a particular path.  My only advice is to keep your mind open, and try to interact with as many interesting people as you can from different fields.  Life is random, but you can definitely stack the odds... :)

Wednesday, August 14, 2013

Essay on Marshall and Paul's SNP FISH paper

(Reposted from PubChase author essay)

An "overnight" success:

Sometimes an idea just works. It's SOOO satisfying when that happens! Whenever it does, I almost feel like it was luck or cheating. But then I think about it a little bit more, and usually it turns out the ideas and the execution have their origins in a deep knowledge acquired through years of work. That's totally what happened with our SNP FISH paper.

Our paper is really simple. Basically, we describe a method that allows us to discriminate single base changes on individual RNA molecules through fluorescence in situ hybridization. It's totally science fiction! But it basically boils down to a couple simple tricks. First, you use a "mask" on your detection probe to block off part of the sequence to increase probe specificity. Then, you use co-localization with "guide" probes that locate the target RNA precisely, allowing one to easily discriminate legit binding from non-specific binding. Sounds simple!

The history of this project is that Marshall (who's a real gear head when it comes to technical stuff) was very interested in coming up with a way to reduce background in our conventional single molecule RNA FISH. He had tried all kinds of things over the span of a couple years, and in the course of doing so, had been working with mask oligonucleotides. He developed all sorts of algorithms for probe design and tried out many different combinations of oligo designs, but in the end, we realized that we needed some sort of way to quickly and effectively test oligo hybridization. At some point, we realized that the best way to do this was to watch a single oligo bind and measure its hybridization efficiency. And this worked, which was great (although it required endless messing around with various dye chemistries, yuck!). So Marshall kept plugging away at that for a spell, when one day he came in, came up to the whiteboard, and said something like "Check this out, I think we can do SNP detection...". It was a really cool moment, because it was one of those rare instances when you have an idea and you just know it will work deep down in your gut. It was also a really special moment for me as a PI, because it was Marshall's idea, and I was so very proud of him for coming up with it.

And work it did, pretty much right away. We basically verified that the principle was solid within something like a couple weeks, then spent the next several months doing the usual prettification that you need for a glossy paper (with huge statistical help from Paul Ginart, who is carrying the SNP FISH torch forward in the lab). But while it seemed to come together so quickly, when I think back on it, it was all of Marshall's years of super technical work with mask probes and dye combinations and so forth that made it happen. I guess the lesson is that you have to work hard all your life to become an overnight success.

Sunday, August 11, 2013

The cost of a biomedical research paper

The science in your average biomedical research paper costs between $300-$500 thousand dollars to produce. I don’t know about you, but those numbers made my eyeballs bulge the first time I heard them. Then I did a little mental accounting, and I realized that number is about right. For a paper in Nature or Science (among the top journals in the field), that works out to around $100K a page! Which raises the obvious question: are papers worth it?

I should start by saying that I am an assistant professor at a research university in the life sciences, and published scientific research is the currency of my profession. I love science, and I believe that we as a society are better for having it in our lives. In a world of unlimited resources, I would say that we should pursue as much science of all kinds as people have interest in doing. But we live in a world with limited resources, and given the realities of government budget cuts for science, we obviously have to make choices.

First off, where is all this money going? As most people who run biomedical research labs would tell you, the majority (if not the vast majority) of the money for a paper goes to paying for the people performing the research. This means salary and tuition for graduate students and postdocs, along with benefits. It is pretty much impossible to lower these costs any further, because they are already so low. Graduate students get paid nothing, postdocs get paid next to nothing, and most faculty could probably increase their salary by 50-100% if they went to the private sector. These people are an employer’s dream–they are formidably talented and will typically work 50-80 hours a week for peanuts. In fact, I remember a graduate student I knew who calculated his hourly wage to be around $3 an hour. Such research, when done in an industry setting, typically costs much more and uses far more robots, which makes it a different kind of research. Oh, and in case you’re wondering, we do it for love.

The rest typically goes to materials and supplies. This can range from 10-30% of the project’s cost, depending on the type of research. Now if you want to make money in science, I would definitely get into the scientific supply business. For various reasons, we end up slaves to various supply companies like Fisher Scientific that charge exorbitant amounts for supplies. Take, for example, aluminum foil. You can get a 12"x25' roll for around $3-4 at the grocery store. At Fisher, it costs $6.50. Lab notebooks cost $40, refrigerators cost 2x what they cost consumers. And it’s not a special “research” fridge, no matter what they say–they often still have the butter drawer in them! Down the line, science stuff just costs a lot, and biomedical inflation is much higher than inflation in general. It should cost much less, but again, as a fraction of the total cost of a paper’s worth of science, it’s not a huge fraction.

Anyway, overall, it’s going to be hard to reduce costs much. So what do we get for these papers? That’s where things get considerably more nebulous and hard to measure, since it fundamentally comes down to how we value science. That said, here are a few relatively concrete observations on the matter. Every week, I get many e-mails with the table of contents of various journals. Of all the paper titles I see, it’s only a small fraction that I imagine more than a handful of scientists would find interesting. Of those, it’s an even smaller fraction that I will find interesting. And then there’s the likelihood that at least 75% of all of these papers are wrong anyway. Now, you might say that at least someone, somewhere would find those papers interesting, even if I don’t. Amazing thing is, though, that a huge proportion of papers are never even cited once. It’s probably hard to get a hard number on this, but I’ve heard estimates of around 50%. And then there’s another huge percentage of papers that get just a few citations in their lifetime, perhaps mostly just from the lab’s subsequent papers. I bet that the majority of papers ever written only get read by the authors (and probably not all the authors) and the peer reviewers (and probably not all the reviewers, either).

Of course, not all this stuff is useless, and citations are not the only way to measure impact. Some papers only show their true worth many years after the fact. Some collect attract huge numbers of citations for a short period, only to be ultimately discarded by the rigors of time. One of the things I’m most proud of developing in the course of my work is a method for detecting RNA molecules in single cells that many other researchers have adopted, and it’s even been commercialized. Now that you can buy it, people don’t cite the paper as much anymore, but that’s okay with me–I feel like I’ve really made a difference, which is surprisingly rare in our line of work. (I believe I got lucky in this regard, by the way…)

Which of course raises another point. No matter how you dice it, the fact remains that a lot of research is just useless. Yes, one never knows what little piece of information could be the key in the future, but some of this stuff just plain doesn’t matter, not now, not ever. It doesn’t get cited because either it’s derivative, it’s arcane, or it’s just plain boring.

Why, then, do we waste our precious research dollars on this work? I think the answer is that it’s precisely because these dollars are so precious. There are special mechanisms for getting high-risk/reward science funded, but for most, getting funding for your research these days means that every single scientific i must be dotted and t must be crossed. Which means you’re proposing to do something pretty boring, most likely. Only two ways out of this: either more funding or fewer faculty members. I think the former is unlikely, so we better brace for the latter.

But I would also say that getting embroiled in metrics like citations per dollar and the such are missing another broader point. Most researchers work in institutes of learning, like universities. This means that our goal is educating people. Unlike most jobs, I expect those who work with me to leave eventually–indeed, I hope they leave and do well. If I have a student who writes a couple papers to get their PhD that never get a single citation, on some level, that’s fine. I would like to think that learning how to do research is a valuable investment in our future. By the numbers, very few graduate students will end up in academia, nor would I expect them to. Rather, the hope is that they will use their training out in the world to do the things that only those with their highly developed skills can do. During their time in academic science, every student and postdoc will lay their own brick on the foundation of knowledge, and some of those bricks will be large and some will be small. The point is that we’re investing in bricklayers, not just bricks.

Friday, August 9, 2013

medium.com is a cool website

My friend Tyler Neylon (who started the website for boycotting Elsevier, incidentally) works at a website called medium.com, I think from the folks who started Twitter.  Their idea is to make a place where you can post longer-form content and have it look beautiful and reach a wide audience.  I cross-posted something from this blog on there, and made a collection for scientific content.  The website is still under development and so is invite only for content posting, but they have a lot of interesting stuff on there–check it out!

Sunday, August 4, 2013

Where are all the special people?

As some of you kind rajlab blog readers may know, I used to be in a band in college. We were a rock band, and I played the keyboards. I wasn't really very good. I guess playing a bit of classical piano as a high school nerd doesn't fully translate to being a rock star, somehow. But it was a ton of fun, and even though it never went anywhere (for me, at least), I consider myself lucky to have been friends with Miguel Mendez, who was the guy who played guitar, sang, and wrote most of the songs. Miguel is a genius songwriter, his lyrics in particular. His songs had that special genius quality that you just know when you hear, and they even if they weren't good they were always great, if you know what I mean (sort of like Martin Scorcese movies). You can find some small fraction of his music online, but he has tons of other recordings and songs that never to my knowledge even got recorded. Some of my favorites I think he only ever just played live for a few friends, like "Dostoyevsky, Nietzsche, Kant, Oi Oi Oi!". Many of these songs still randomly pop into my head now almost 15 years later.

Miguel was from Long Beach (same high school as Snoop Dogg), and one of my fondest memories with Miguel was when I went down to visit him in Long Beach one summer when I was in college. He was staying at his mom's house and not doing much in particular. It was pretty hard for him to get around also, because he (artist stereotype alert!) is an epileptic and so he wasn't technically allowed to drive. (One of my favorite Miguel sayings was "Living in Manhattan without money is like living in LA without a car.") We basically just hung out for a few days or a week or so. He showed me around the burrito joints and stuff, although I can't really remember all that much of it. One thing I remember was meeting up with one of his friends in her mom's mansion in Santa Monica, with her mom bombed out on Prozac on the couch–I was thinking "didn't I see this in a movie somewhere?" Oh, and I remember Miguel getting me a pair of uncomfortable orange shoes at the grocery store for $5.

But otherwise, we just sat around the house. He showed me some of his old 4-tracks that he made music on when he was a kid, when he and his friend Farmer Dave used to try and show each other up with their latest awesome song. We talked about our rock and roll aspirations, which I had bought into at the time. Miguel said that he wasn't afraid of being a rock star, and of the "moments of bad poetry" that went with it, as he put it. He also said that he always know that he was special, from a very young age; everyone told him so and he just knew it, on the inside. It was just a matter of which direction it would take when it expressed itself. I'm glad it took the form of music. When I was in Long Beach, Miguel gave me a tape of his "Machaca Beef" album, which was just him goofing around on a 4 track one summer at his mom's. I'm pretty sure I've lost the tape now after all these years, but I really would love to find it again–it is still one of my favorites.

After college, we all moved to NYC, and I eventually lost touch with Miguel and the band for various reasons. I sometimes wonder where he is and what he's doing. I'm pretty sure he's still in New York making music. And then, the other day, I was wondering where all the remarkable people I met in college were. I met a number of them back then. Some of them were, like Miguel, great at music. One guy was an amazing cook. Another guy, Jamie, was an incredible dancer and last I heard (which was many years ago), he was going on tour as a dancer with Paul McCartney. And I occasionally met some in class as well. You've probably met those types, too. They typically do math and physics, and are done with college by the time they're 17 or something like that. I ran into a couple in high school, and then again in college. They're not just regular smart, they're so smart that the normal way things are done just don't make any sense anymore. The two I remember from college were in high school at the time (!), and they were both in a graduate level algebra course that I took as a senior (taught by another genius, Robin Hartshorne). One was Gabe Carroll, who I think went on to win the Putnam four times or something crazy like that–I didn't really talk to him at all, but I could easily tell his mathematical abilities were absolutely astounding. He would often ask questions in class that nobody but Prof. Hartshorne could even understand, and judging by the professor's response, they were clearly substantial and insightful.

The other math supergenius I met in college was this kid named Dave (I can't remember his last name, or whether I ever knew it to begin with). Dave was a quiet and unassuming guy who would just sit in the back of class and work out the math on a sheet of paper or in his head. I don't think he even finished the class–probably found it a waste of time. I think I remember "working on math" with Dave once. At around the same time of this algebra class, Dave and I were both sitting in on a differential geometry course taught by Charles Pugh. It was insanely hard, and the homework problems were from this book by Hirsch, which I must still have lying around somewhere. The problems were so hard that I couldn't really understand what the problems were even about most of the time. Dave and I sat in the math lounge one time, and I was furiously writing and talking about my thoughts. Dave was just sitting there, daydreaming and occasionally responding to what I was saying. Whatever he said was so nice and encouraging, sort of like how you tell your kids "good job!" when they've added 43 and 89. But it was clear that whatever thought I was laboring on now, his mind had already seen long ago, and that his thoughts were going off in new places, exploring ideas in a split second that I, sadly, would likely never experience myself. I imagine the connections he could make were quite beautiful, like a symmetry that only becomes clear once you have glimpsed both sides.

More often, though, I would run into Dave at the little alleyway that is Durant Food Court, aka the "Asian Ghetto", where I would often go in the late evening to get some food (Vietnam Village was my favorite). We would talk about all kinds of things, often of the more philosophical nature that was my tendency at the time. It was very interesting to talk with Dave about his mathematical abilities. He had a remarkable ability to never make you feel worthless, but he also didn't bore you with false modesty, like "Oh, I just took advantage of some opportunities when I was young" or whatever other bromides you often hear from gifted people. He spoke quite plainly about the fact that he had a rare mathematical talent without the slightest hint of braggadocio, which meant that we could quickly move on from the dull "Yes you are, no I'm not" exchanges and instead talk about more interesting things. Once, I remember we were talking about the creative process, comparing in particular math and music. I was saying they there were some inherent differences, while Dave was arguing that they were deeply, fundamentally the same. To which I replied "Math has never made me cry." Dave said "That's true... that is a qualitative difference..." and sat thinking for a while. One of the great things about Dave was that I knew he wasn't thinking about how to out-argue me on this point (it would have been easy for him to "win"), but rather was considering what this meant, and how it might change how he thought about something. Dave was in many ways different than a lot of math whiz-kids in that he seemed much more interested in thinking about the world than in showing he could solve harder problems faster and better than you could. I wish I could remember more of our conversations.

One of the interesting things about these super math genius types is that they often end up in strange social juxtapositions due to their age. For instance: one of the people through whom I got to know Dave was this tall math/physics/computer science guy named Noah. Noah was an interesting fellow, who always had a story to tell, often involving experiences with psychedelic drugs (it was Berkeley). Noah told me that one time, Dave came to his house and wanted to smoke pot with Noah. Noah happily obliged, but pretty soon Dave just got way to high. So then Noah said, "Do you want me to drive you home?" To which Dave said "No way, I'm way to messed up. Hang on, I'll call my friend and see if I can stay there." So he called his friend and got the okay, and Noah drove him over there. Noah knocked on the door, and... his computer science professor answered! Noah was like "Uh... I'm here to drop off Dave...", and the CS professor also looked at him awkwardly, said "okay" and they both just pretended like it never happened. Another time, I had Dave come to one of our music shows (you know, from the band I was in). I think he enjoyed himself, and we were all hanging out together outside when Dave's mom came to pick him up, which felt somewhat incongruous to say the least. Jamie (the dancer) was there, and said super loudly "Oh, your MOM is here, oh that's just so cute." It was really obnoxious, and I think it really hurt Dave's feelings, and to this day I regret not having told Jamie off about it. I guess it will have to remain a confrontation imagined rather than experienced, as is often the case.

The thing I've been wondering about lately is why I don't encounter more of these people in my professional life. Where are they all? Seems like there are a large number of very smart kids, and yet the vast majority of the faculty I meet seem to have gone through the normal academic path at the normal academic pace (myself included). I think I know a couple people who were the prodigy types who are now faculty members somewhere, but now they just seem normal. And that's when I realized that the reason I don't notice all the smart kids is that I'm surrounded by them! It's just that after so long, you get acclimated to them, and since you're all doing the same thing and working on the same problems, you just don't notice it anymore. People like Miguel and Dave always fascinated me because I could never understand how they were so good at what they were good at. But Miguel was always hanging out with other musicians who could understand, like his friend Joel Morales, who was also awesome at music. (Miguel used to say that the reason he thought Joel was better than him was that when people listened to Miguel's music, people would say "How did you ever think of that?", whereas Joel's music would make them say "Why didn't I think of that?". Same applies in science a lot of the time.) And the halls of math departments actually do house a number of former math prodigies. Meanwhile, I feel like the people in my lab are awesome at what they do, and do work that very few other people in the world could. It's just that we do it together, every day, all day, and so perhaps we (or at least I) take it for granted that we had to be special (and lucky) to get here and that we're fortunate to be around so many talented people.

Some of the prodigies, however, disappear and are never heard from again. Some burn out, and some who aim for academia can't make the transition from problem solving to research, which are two different things entirely. I think some end up using their "gift" in whatever way interests them most, which is of course all for the best–for instance, last I heard of one guy, he was really working hard on fully optimizing his home theater PC with some open source TV tuner. As for Dave, the very last time I saw him was at the food court, and he said he was going into the Peace Corps for a couple years. Wherever he is, I wish him well.