Monthly Archives: July 2007

Farewell to…

Odile Crick, wife of Francis. Interestingly, the illustrator of the figure depicting the DNA double helix in that Nature paper:

In his memoir, “What Mad Pursuit,” Dr. Crick recalled going home that day and telling his wife of the historic discovery. Only years later, he wrote, had Mrs. Crick told him that she did not believe a word of it, saying, “You were always coming home and saying things like that, so naturally I thought nothing of it.”

Bio::Blogs #13 call for submissions

It’s almost time for the monthly compendium of bioinformatics-oriented material from the blogosphere that we call Bio::Blogs. I’ll be hosting this month so send your submissions to the usual bioblogs <at> gmail <dot> com. Remember, you can send one of your own posts or recommend someone else’s.

The official publishing date is August 1st (Wednesday), but I may delay until Friday 3rd, depending on other commitments this week.

A nasty MOD_RES surprise

A lot of bioinformatics consists of fetching files in various formats from databases and writing parsers to extract features. What to do when one of your trusty parsers unexpectedly fails?

  1. Don’t panic
  2. Make sure that you haven’t done something silly:
    • did you inadvertently alter the code recently?
    • did you run a different version of the code by mistake?
    • did you use the correct file(s) as input?
    • does the machine that you’re using have the required libraries and software used by the parser?
  3. If your code or machine setup hasn’t changed, then the culprit must be the input file

Take a look at the file – use something like grep if possible to examine specific lines and see if their format has altered.

One of my more robust perl scripts is designed to examine the MOD_RES line in the feature table section of a SwissProt file for protein kinase names. Imagine my surprise when out of the blue, not a single name appeared in the ~50 000 line output file. A quick “grep MOD_RES file.dat | less” revealed this alteration:

Previous:
FT   MOD_RES     353    353       Phosphoserine (by MAPK12 and MAPK9)

Current:
FT   MOD_RES     353    353       Phosphoserine; by MAPK12 and MAPK9.

Might be time to fix up your regexes if you have code that parses SwissProt format.

HOWTO: generate random sequences using EMBOSS and Perl

Once in a while, you need a “random” sequence.  For instance, you might perform a simple statistical analysis on a bunch of sequences from a database and decide that it looks interesting.  However, if the analysis looks the same when performed on a set of completely random sequences, it’s a fair bet that it isn’t interesting at all.

Read the rest. . .

Finally: ATI fglrx + xgl + beryl for Ubuntu feisty

I’ve had harsh words to say about ATI Linux video drivers in the past – and I’m not withdrawing them. My advice is still do yourself a favour and buy NVIDIA. That said, I did an Ubuntu install on a work machine today (we get little choice with the specs) and finally got the proprietary fglrx driver working with Xgl and beryl, for a fancy desktop. All thanks to this very clear post; I’ve read many and this is the best.

The key points are:

  • The open-source radeon driver has composite (for beryl/compiz), but no TV-out and doesn’t support newer cards
  • The closed fglrx driver has TV-out but no composite. However, you can get beryl/compiz to run by combining it with Xgl, which is as simple as installing xserver-xgl and editing a small start-up script.
  • Here’s the key: the beryl packages in feisty are version 0.21 and don’t like Xgl. You need to downgrade to version 0.20 in the beryl project repository.

Xgl can do nasty things to your fonts (this thread includes a fix for “my emacs is all squares”) and some apps, like Google Earth are difficult to run (this page contains a fix) but on the whole, things work. If you’re stuck with ATI give these tips a try.

One day either fglrx will have composite or radeon will have tv-out. Which first, I wonder?