“I am thinking now”

When you're an educator, it's easy to get bogged down in details. Often petty details. And it's easy to forget the big picture.

In this TED talk, Patrick Awuah talks about how universities create the educated people that can drive nations to succeed. And how the lack of such people can cause national crises.



He calls this leadership, but I'm suspicious of that word.

Just another day in the lab

Courtesy of the Simpsonizer.

Cloudburst!

Lots of lightning and raucous thunder late this afternoon. Internet access at uni dropped out. The usual flood points were, as expected, flooded. I can make it almost home with only the souls of my feet getting wet... but there's a couple of ankle deep spots near the end of my trek. Alas.

Meanwhile, spent a good chunk of the afternoon – before everyone started to worry about being smitten with Thor's hammer – troubleshooting some DNA sequencing. It looks like we simply had too much DNA. Yes, for our delicate sequencer, it is possible to have too much of a good thing.

The sequencer is broken again, though, and so we won't have a chance to test our suspicions until Wednesday, probably.

Revision of a figure

Yesterday, I did a quick revamp on our Neuroethology poster for HESTEC. Previously, I wrote about the design of one of the data figures for that poster. Interestingly, I found out at the meeting that I was too clever for my own good in designing the Neuroethology poster. The problem was that I used the bars coming from the boxes to represent the minimum and maximum. Several people interpreted these as error bars (standard deviation or standard error). Because overlapping error bars usually indicate that groups are not significantly different, and error bars are smaller than minimums and maximums, this led some viewers to momentarily question the results.

Consequently, I went back to a more standard bar graph in the revised HESTEC poster. It's similar to this one, except it shows transformed data rather than the raw data, and the colour is not bright red.

The size of the HESTEC poster was also smaller, so a lot of cutting text and general simplification occurred. It's probably a better poster as a result.

I also started work on my annual compilation of everything I've done in the last year. It's a dreary process, although sometimes it can be nice to see how much you've done.

Getting closer...

It may not look like much. And it isn't. Nevertheless, it's my first little steps into a new area of research: DNA analysis.

The traces above show a small snippet of an attempt to sequence a bit of DNA. Ideally, there should be a nice sequence of evenly distributed peaks of about the same height. Each peak indicates a single nucleotide. Unfortunately, our sample is not ideal and there's a lot of noise in the measurement. My student Unnam and I need to do some troubleshooting and try again.

Another indication of our problems is that DNA is made of a sequence of four letters: A, T, C, and G (nucleotides, really, but the names need not concern us here, and the letter symbols are universally used). The sequencer gave us A, T, C, G... and N. "N" is a symbol for "any nucleotide," which basically means that the machine couldn't figure out what it was and spit out a "I dunno."

We were very excited to be sequencing -- finally -- yesterday, so that it was really not all that usable was disappointing. Or, as Unnam wrote, "THAT SUCKS. A lot."

Still, after my buddy Virginia was always bugging me... "Zen, you have to get into molecular biology to show the world that you're a modern biologist," I do feel the need to point out that I'm moving in that direction.

Omit needless words

Lately, I've been seeing a lot of headlines that say things like, “10 places to visit before you die.”

Is there really any other time to visit those places? Does anyone think, “Oh, that's okay, I'll get to that after I die...”?

A rule

Nature doesn't abhor a vacuum.

Nature abhors a contradiction.

Busted

DNA sequencer: $100,000.

Electron microscope: Tens of thousands of dollars.

Floor centrifuge: Tens of thousands of dollars.

The feeling you get when all of them break in the space of a week or so: Priceless.

I swear, this place was built on an ancient Indian burial ground. Sometimes I think a curse is the only explanation.

One of my students has been working for over a month to get DNA samples ready to sequence, and came in this morning only to learn the sequence broke Wednesday night.

Art and science

I was talking to an artist yesterday who wanted to incorporate some microscopic images into a new pork she is preparing. This got me thinking about the similarities between art and science. Here's a few.

Both are profoundly creative endeavors.

“The overriding message of all art... is ‘Pay attention.’” - Harlan Ellison. You could say the same thing about science: the message is, “Pay attention!”

Both are hand crafted, not mass produced. At least, in their original form.

Both, at their best, transform the way you see your world.

Headspace

I was sick on Saturday, had an uncomfortable overnight flight back to Texas on Sunday, have had the south Texas heat kicking my ass every time I walk out the door, and am still sleeping in way too late.

My head is not yet back in the game.

Some of the routine paperwork is done now, but that's about it.

Champions never quit

Brian May, of the legendary band Queen, has finished writing his doctoral thesis. Thirty years after he started it.

Now that would be a graduation ceremony to see...

Scene at customs

"Edinburg," says our customs officer in the Vancouver airport. "Who lives there?"

"There's a university there," I say.

"Pan Am," he says, which makes my eyebrows crawl up my forehead. Nobody knows where Pan Am is...

"I'm from McAllen," he says.

"What?!"

Much better, though, than the flight attendant on the flight from LAX who kept calling McAllen, "McLean."

ICN 8 continued

Museum of Anthropology rocks.

So many delegates leaving on Thursday night when the conference ends Friday afternoon sucks.

ICN8 (with news on 9)

Vancouver Aquarium rocks. The beluga whales really are hypnotic. And sea otters are pretty darn cute. And so much more.

Fireworks rock. Went on a dinner cruise and stayed in Vancouver harbour to watch Spain's entry in an international fireworks competition. Nice big explosions.

ICN9 will be in Spain in 2010.

ICN8

I am blogging (if this works) from my Pocket PC in Vancouver, BC, where I'm at the 8th International Congress for Neuroethology. The sun came out, the science is good, and I gave a short "ad hoc" talk.

And I have unlimited access to Canadian chocolate.

Oh yeah. Lovin' ICN8.

ICN8

Is in Vancouver.

_________________________________________________________________
http://liveearth.msn.com

See ya later, south Texas

Tomorrow morning I get on a plane for the Eighth International Congress of Neuroethology. One of my favourite meetings, but only held every three years, and I've missed the last two. 2001: Moving to start new job in south Texas. 2004: Had world-class experts down in region to help with research.

Right. Time to sleep to be awake enough to catch the morning flight.

Poster done

I just finished printing off my poster on the printer so big, they call it "Tank." Fortunately, as it was printing, I only saw about 3 things that I would change. One was a colour issue: some data I had printed in blue came out a bit darker on print than it does on screen, so the data points were not as distinct as I would have liked. A second was that I probably could have used some colour in the graph I described in my last post to make the mean diamond stand out a bit more. Third, some of the spacing between text and heading could be improved.

But if those are all I'd tweak at this point, I'm doing better than average, I reckon.

Making of a figure

It's late. Stupid late. But I just finished up a poster for a meeting. The last thing I was doing were some statistics. I had to do them by hand, because they're specialized enough that most computer stats programs don't calculate automatically.

But I thought I would share the evolution of one of the figures that went into the poster.

First thing I do is just plot the raw data, shown below. This is just for myself, not for presentation (besides this "behind the scenes" post in my blog, naturally), which is why it looks pretty poor. I don't like the bars in red stripes.

This is a useful step just to get a sense of what you've got, and sometimes helps detect errors. I found one of my students misplaced a decimal this way, so one of his data points was out by a factor of ten, which wasn't good. But we caught it.



That's the data from one experimental treatment. Now I want to see all the experimental treatments side by side. This one I was thinking I might end up using in a presentation at some point, so I cleaned it up a little more.



The good news is that it looks like there might be an effect. The bad news is, that from looking at the plot above, the data are not normally distributed -- most are piled up over on the left hand side -- and they differ in how much they vary. Both of these things are bad statistically.

The plot below shows how I transformed the data to try to fix those issues.



Not perfect, but certainly not as skewed toward the right as before. Again, this is just for my own exploration, so it's just the default red stripes. I could change the default, but I've been too lazy.

The next step is to run some stats. Here, I have to switch to a real stats program, which does all the test right -- but leaves a lot to be desired in terms of graphs.



The above was the default plot of averages I got when I ran the statistical test -- which confirmed that there was a significant effect! Still, the plot leaves much to be desired. I want the data points in different order, and I don't want them joined by lines, and I want to show some measure of the variation.

The stats program gave me this when I asked it to show mean and standard error.



Still not great. And the above two pictures are both screen grabs. I want an image in a form that will scale up and not get all jaggy when I put in on a big poster. The WMF format scales up, but when I try to export the graph in WMF, I get this:



Proof that the stats software is about the numbers, not the pictures.

So now I go back to my graphing program. I try using it to plot averages and error bars, and get a a basic bar graph.



Not bad, but because one of the issues with this was initially the skew and variation, I want something that might show a little more detail than that. I try a box plot of the raw (i.e., non-transformed) data.



Getting closer now, but the fiddling over details gets more intense. The lines are too thin for a poster, so I thicken them. The little square in the middle, which shows the average, tends to get lost when I thicken the lines; I turn that into a diamond to make it more distinct. The top and bottom whiskers are supposed to represent the 95% confidence intervals, but the sample size is small enough that it ends up being in the same position as the minimum and maximum, which are shown as the top and bottom Xs. So I get rid of the Xs while I'm at it.



Now I also want to show the transformed data, so I make a similar plot and change it much like I changed the graph of the raw data.



I add the letters above each box to show which condition is statistically different from the others. (Boxes that have the same letter above them do not differ.) But the plotting program doesn't allow me to line up the letters as precisely to the boxes as I want, so I import that into a real graphics program for final tweaking.



And that's the end result on the poster!

To make this one figure, I used four different software packages. Microsoft Excel 2003 for data manipulation, Origin 7 for graphing, SPSS 12 for statistics, and CorelDRAW 12 for final touch-up. I'm showing almost a dozen graphs, although there are a few more steps in the process that I didn't show here.

And this is the easy figure on the poster.

Skeptic or denier?

In my earlier post about lying with statistics, I mentioned scientific skeptics. Part of what I was thinking of was a recent (30 June 2007) edition of The Science Show about climate change (again), which included comments by Ian Plimer, which is documented in more detail in a following In Conversation episode (5 July 2007). It looks like climate is also on the agenda for this week's episode, which I haven't listened to yet.

Also saw a recent TED talk by economist Emily Oster about AIDS in Africa that challenges a lot of ideas about the disease.

That got me thinking about what's the difference between a researcher taking a minority view compared to someone who is just in denial? Who's a thoughtful doubting scientist who is thinking seriously about evidence versus someone the flat earther? (Setting myself up for hate mail from the flat earthers... yes, there really are still people out there who believe the earth is flat.)

I think there are a few guidelines.

First, is the skeptic willing to admit that he or she could be wrong? What would it take to convince you that you are wrong? If you're concerned about evidence, what evidence do you want? Is there an experiment you'd suggest should be done? What numbers do you need to see? (Plimer, for instance, expresses doubt that humans are causing climate change, but he doesn't really say what he get him to be convinced. Is there anything that would convince him or not? I don't know.)

Second, is the skeptic actually doing active research on the topic? It's one thing to read a lot of papers. Reading a lot of papers is surely important. Nevertheless, it's another to actually do it. And when you look at a lot of cases of prominent scientists who take minority views, they are often on matters outside their actual research. Lynn Margulis and Kary Mullis, for instance, are both well known scientists who expressed doubt that HIV causes AIDS, but neither of them practising virologists or epidemiologists.

And of course, the third is just how small is that minority view? This one is often really hard for outsiders to judge, particularly because media coverage is notorious for presenting opposing views as though each has equal claim to legitimacy. Someone sees a show with one person saying X and the other saying Y. There's often no easy way to tell if position X is pretty much subscribed to by researchers in every university in every country of the world who are funded through competitive grants, while position Y is backed by advocates numbering in the single digits who are backed by a private foundation.