Uncategorized

A few weeks ago I was reading an article which claimed before the recession seven times as many PhDs were awarded in the biological sciences as there were openings in tenure track positions. Of course in between finishing grad school comes years of post-doc work, but in the end PhDs in must equal PhDs out.

So assuming every PhD graduate wants to be a professor (probably not true) that means even after making it past admissions committees and qualifying exams and thesis defenses, these newly minted PhDs face an 86% washout rate in their quest for a faculty position.

Eighty-seven percent. Let’s put that in context. These are the numbers I turned up with some quick googling:

• Roughly 10% of marine recruits drop out during basic training
• Roughly 55% of people going through the training to become army rangers drop out
• In an average year 70% of the people who start training to be Navy Seals (the folks they sent in when they finally found Osama bin Laden) don’t make it to the end.
• To actually find a training regime with a higher dropout rate than the road from PhD to Professor I had to go to the wikipedia page of the Pararescue Jumpers — the guys who jump out of the rescue helicopters into enemy territory to rescue the wounded. Their washout rate in 90%.

It’s hard work keeping up with a blog while being a grad student, but some people find the time to manage it.

Here’s a new site: ProSeed with Science written by Greg, the same guy who used to write Pie-ence. First few posts look interesting (lots of sunflower stuff, a crop with a genome even bigger than that of maize!), and he’s already made it past the three-post line before which most new blogs die a quick death of neglect. Check it out!

QTL stands for quantitative trait locus. Which raises even more questions. What is a quantitative trait? What is a locus?.

• A locus is simply a region within a genome. Anything from a part of a single gene to a large hunk of a chromosome.
• A quantitative trait is one where different individuals vary continuously (like height or weight) rather than falling into discrete categories (like whether a person has blue or brown eyes*).

A QTL is simply a part of the genome that has been show (using complicated statistical tests) to influence a quantitative trait like height. For example, people with one particular region of chromosome 8 tend to be slightly thinner than people with other versions.** Now a lot of qualities we’re very interested in as a society turn out to be quantitative traits. I’m not even going to touch the implications for human genetics, but within plant biology lots of the things people are really interested in changes, from flowering time, to drought or disease resistance, to the big kahuna of them all YIELD, are quantitative traits.

How are new QTLs discovered? It’s not as simple as classical genetics where you can simply run a mutant screen, pull out individual that look weird in a way that seems interesting, and identify the gene which was mutated to create the change you observed.*** Instead a researcher has to measure their specific trait in a bunch of individuals (easily done for something like height, less easily done for something like number of root hairs per centimeter of root or trichomes per leaf****) and then compare those measurements to a bunch of information about the genomes of each of those individuals. If the average height of all the individuals with version A of some part of the genome is higher than the average height for all individuals which have version B of that same part of the genome and that difference is significant after a whole bunch of statistical tests, then that region is a QTL.

Do all that and congratulations. You’re done now. You can go publish a paper describing your discovery of QTL controlling whatever trait you just measured! Depending on the species, the trait, and how many (and how small) the QTL you found, that paper could be anywhere from a major finding to something buried in a never-heard-of-the-name-before journal. QTLs are one of those weird case (like cell phones) where smaller is better.

Why? Because the logical next step, after identifying a QTL, is to figure out what it is about that region which influences the measured trait. If the QTL in question is too large, that could mean trying to take a list of dozens or hundreds of genes and, somehow, devising a test to prove: It’s this one! Gene AT1G15210 helps regulate height (or root hairs, or trichomes, or whatever it is being studied).

If you’d like to check out an example of what an actual QTL paper looks like, I really enjoyed a recent one in G3 (G3 is open access, so everyone should be able to access this) at measured the development of tassel like outgroups on the end of maize ears. I’ve run across this a few times in the field back when I did actual maize genetics and always wondered what was going on genetically to create such weird looking plants. I still don’t know for sure, but now I know there are real geneticists out there working to discover the answer:

* Yes, if you’ve spent any length of time staring into a number of women’s eyes (or men’s for that matter) you’ll know there’s a great deal of variation within those categories, but the point is there ARE obvious categories for eye color, while any attempt to group people by weight or height would rely on essentially arbitrary cut offs.

**This statement is used as example. I know absolutely NOTHING about human genetics. You have been warned!

***And to be honest, in practice there is nothing simple about classical genetics. I’ve been forcefully reminded of this in an ongoing e-mail discussion that has gotten into the long term pedigrees of individual maize seeds.

**** I’ve often wondered if grad students assigned to such QTL projects have significantly higher than average drop out rates.

Read the whole list here, but I want to highlight #8 in particular:

Evolution has a requirement that things work, not that it’s an elegant engineering solution. Expect jury rigged systems which can be bewildering in their complexity.

This was just on slashdot, so I imagine many will have already read about it, but for those who haven’t, here’s a wonderful metaphor to understand the difference between how scientists (biologists anyway) code, and how professional computer people (some of whom as also scientists) do:

Scientists see their software as a kind of exoskeleton, an extension of themselves. … The software may do heavy lifting, but the scientists remain actively involved in its use. The software is a tool, not a self-contained product.

Programmers see their software as something they will hand over to someone else, more like building a robot than an exoskeleton. Programmers believe it’s their job to encapsulate intelligence in software. If users have to depend on programmers after the software is written, the programmers didn’t finish their job.

The full post was writing by a fellow named John D. Cook and is available over on his website.  (more…)

Papayas genetically engineered to resist papaya ringspot virus were developed in the public/non-profit sector (Cornell University, the University of Hawaii and the US Department of Agriculture). So none of the “I don’t have a problem with the science of genetic engineering, I just don’t like/trust big companies patenting life” arguments apply.

The engineered papayas were released as the ringspot virus began to devastate the Hawaiian papaya crop back in 1998 and have been grown and consumed successfully ever since.

The engineered papayas’ resistance is the result of expressing a protein from the coat of the papaya ringspot virus and engineered papayas contain less of this protein than the fruit of infected trees. Yet the fruit of diseased trees can be sold as “organic” while the fruit of healthy resistant trees (distinguished only by containing less viral protein) cannot.

The engineered papayas even provide herd immunity that makes it possible to grow un-improved organic papayas for export to countries like Japan that reject much genetic engineering.

But the engineered papayas do have one clear (and sometimes fatal) flaw which is only now becoming apparent. They aren’t immune to the machetes of the ignorant.

h/t @Franknfood

I was very excited to read in my twitter feed this morning that google has launched a new service that lets researchers automatically aggregate data on all the papers they’ve published and how often those papers are cited. With a click of a button you can opt in to sharing that data with the world (or at least anyone who searches for your name in google scholar).

Since we’re always told its important to judge the quality of researchers by the impact of their papers (presumably measured by citations or more advanced metrics like the H-index) rather than the impact factor of the journals their papers are published in, I think this represents a big step forward for three reasons:

• Unlike other services that do sort of similar things. Google Scholar profiles are free and visible to anyone (once you opt in).
• The service automatically updates as newly published papers cite your previous work, and as you publish new papers yourself (no need to remember to visit your page and manually enter each new paper). People who have browsed faculty profiles on university websites and realized the professor hasn’t remembered to update their list of new publications in five years or more will appreciate now important a feature this is!
• It’s surprisingly accurate! The only corrections I had to make to my profile were condensing two duplicates of existing papers which were listed with slightly different titles or author lists on different websites. Google scholar didn’t miss a single one of my papers, nor did it include any of the papers published by other people what shared my name back in the 20th century, like so many other searches have.

For those of you who don’t dive into the comments section of posts, I wanted to share some really good points others made commenting on my previous entry:

William Nelson points out that whatever choice you make for computation support (collaboration/hiring in house/farming it out) it makes sense to involve them from the initial design phase, rather than coming to them with a bunch of pre-generated data. Generating a dataset that will play nice with computational pipelines often means an experiment that has a stronger overall design, so this is really a win-win for all involved. The same logic extends to designing follow up analyses. Explain what you’re interested in finding out, not just the specific analysis you’d like run. There may be much more effective or faster ways to find out the answer to your question that someone familiar with the nuts and bolts of computational work can suggest if they understand the whole problem.

He also very correctly points out that as expensive as hiring computational biologists or programmers who understand biology can be, people with those same skill sets make a lot more money outside of academia. So keep your programmers (whether co-workers, employees, or collaborators) feeling happy and valued! (A woman who entered grad school the same time as I did brings her lab’s programmer cookies each time she needs something done, and she’s getting papers so fast she is on schedule to graduate before anyone else in our incoming class. )

Meanwhile Matt (the scientist gardener) reminds everyone that being a whiz with computers and a whiz with statistics don’t go hand in hand. And for big data projects like most high throughput sequencing experiments, a lack of statistical expertise can hamstring a project just as effectively as a lack of computational skill. So your university’s statistics department is another set of intelligent colleagues you should remember to develop and maintain good relationships with.

He also reminds us that computational analysis isn’t the only talent people can use to get through a PhD without developing the skills necessary to direct a research project of their own as a Principle Investigator later in life. It’s all too easy to fall into the same trap by being the one person in lab who is good at some complicated molecular biology technique. The example he used was chromosome walking… which I’m ashamed to admit I had to look up on wikipedia.

Under absolutely no circumstances should you take your hard drive full of data, walk into lab and drop it on the desk of some new grad student who decided to go to grad school because he loves plants (or whatever your favorite model organism is) and was a wiz at PCRs in his undergrad lab and tell him he’s now in charge of figuring out how to turn it into a paper.  (more…)

“Back in my day,” countless middle aged professors have said, “if you cloned a gene in grad school, that was it, you were done and graduated.”

Well times change, and cloning a gene isn’t quite as hard as it used to be. But don’t let the nostalgia a lot of old school geneticsts give off fool you into thinking identifying the gene responsible for some interesting mutant phenotype isn’t still a big deal.

Here are the three most recent papers I can think of off the top of my head reporting the cloning of maize mutants: