Blind.Scientist

Alerted by a message sent to the EvolDir mailing list I went to check the Scientific Direct Customized Mailing Lists. Very entertaining product.

Apparently is not only in bioinformatics that you can publish Greasemonkey scripts, Excel macros and 40* errors on PubMed entries.

BMJ also has some very useful science and with its impact factor higher than 9, I can see a lot of opportunity to publish there. I guess just this short article alone would be able to “provide a rare opportunity for the chocolate industry to become palpably involved in public health promotion”.

Last week I registered to use a large online database of biological data. So far, so good, I started checking out what I was able to do, how to get data, the whole nine yards. Until I got to the main page, listing most of the database entries. I noticed that Firefox wasn’t showing the vertical scrollbar, so, evidently, I wasn’t able to scroll down and up the page to see the data.

Problems like this happen when the scrollbar is attached to a div instead of being set for the whole page. As a diligent user I dug for the web admin email address and sent a short message with the possible error, including a couple of screenshots, one from “buggy” Firefox and another from Konqueror (which displayed the page with no “errors”). I also mentioned that Firefox under Linux and Windows had problems showing the page.

A couple of days later I receive a polite reply explaining that Firefox isn’t supported to display their pages. Sorry!? I cleared my eyes and then I found a link to the requirements page where I saw a second time that they suggest you to use Internet Explorer 6 or 7. Next thing I did was to check a GPS just to make sure I wasn’t in Punxsutawney.

My humble internet “portal” served by the good guys at GeekISP are accessed by the following browsers
Firefox 49%
Internet Explorer 32%
Safari 10%
Opera 3%

I heard a rumor that it is 2008.

That’s for all the bioinformatics developers that distribute executables with spaces in their names. Thanks.

From this post I became aware of the paper “Alignment Uncertainty and Genomic Analysis” by Wong et al..

We learn valuable lessons, some evident and some implied. I am covering only the evident here, the implied I leave for the imagination of who’s reading the paper. To the evident lessons (conclusions):

• Alignments are important for genomics studies
• Some genes are more difficult to align than others
• Some alignment programs align some genes better than others, while other alignment programs align other genes better than some
• Most genes in an organism are under the same DNA model of substitution, hence they all have identical selective pressure, drift, migration, recombination and mutation rates

There are things you can only learn from Science.

As promised I’m writing about the free version of TNT. So far I was able to test the Linux version which does not have a graphical interface. I am planning to download a Windows version and test it, but not until next week. So far here are my impressions.

When you start TNT you are greeted by its own prompt, in identical fashion to PAUP and MrBayes. As a regular user (or most of the regular users) I fired up TNT without reading the manual, so my first attempt was to write help at the prompt and the prompt answered by changing itself from “>tnt*” to “>help”. One point against TNT, as either PAUP of MrBayes at least display a list of possible commands. Feeling helpless I decided to access the manual.

The Manual

Oh, my eyes!!! Whoever chose the background colour for the html manual might have an agreement with some optometrist. I cannot read a paragraph without averting my eyes for a “refresh” (if you are interested search for bgcolor=aqua). Anyway, the introductory part of the manual tells you to do exactly what I did, by entering “help” at the prompt. Second thing it tells you is that if you don’t have a Windows version it is not possible to convert your files (i.e. fasta, phylip, nexus) to TNT’s own format. Usual, it follows the rule, one application - one format, they just forgot to include one converter. Just adding insult to injure they suggest that you use a text editor, anyone, and check their example files. Yeah, right. Basically TNT read Hennig86 matrices. Just get them from your 5 1/4″ floppies.

update: there is a way to display the commands at the prompt. “help” alone does not work, it has to be followed by a “;”, but the manual says “help “

update II: according to the manual TNT is able to read NEXUS files and can even include TNT commands in a block. Good.

Using TNT

Apparently TNT has a good selection of commands, from what can be seen from its help. It covers most of the things you want to study by using parsimony. It is mentioned in the manual that TNT does not do maximum likelihood (don’t say!) or bayesian analysis (geez!), but they will incorporate some ML algorithms, even though rudimentary (I bet!).

So, from my shell I started TNT and as I would do in the PAUP CLI version just put the filename as a parameter

./tnt primate-mtDNA.nex (if you really noticed I was using one of PAUP’s example files)

No dice. So I dig deeper in the manual and find a section dedicated to us poor Linux and OS X users

Linux and Mac OS X versions of TNT have a couple differences with the standard versions. First, those versions can take a comma ( , ) instead of a semicolon ( ; ). This is because the shells there use semicolons to separate commands, and giving arguments to TNT from the command line becomes difficult. Second difference is that those versions can run in the background. If you want to start TNT in the background, use bground as first argument, then give the normal arguments you would give the program (do not activate reports and do not change the silent status, but you can set report+ to report every certain number of trees, replications, or seconds, so that you can monitor progress by inspecting the output file; make sure you do have output files, otherwise the program would run and exit without doing anything!), and end with the usual ‘&’. TNT will not write to the console, or expect console input (at least if no errors occur). It will write its results to a file and exit smoothly (exit code is 0; if the program runs out of commands while in the background it exits with code 1). Should some error occur during the search, the program writes a message to the ouptut file, and exits with code 255. Try

./tnt bground p zilla.tnt, echo=, log peep, rep+1, mu10=ho3, le, ne, quit, &

and do “tail –f peep” to monitor search progress.

Very simple and straightforward. But at least I got a clue on how to open a file or use it as a parameter. So I repeat my first command line with the parameter p and then the filename

./tnt p primate-mtDNA.nex

Success … no not so fast. An error is reported:

PISH (Phylogenetic Inference SHell)

Reading from primate-mtDNA.nex
Reading file primate-mtDNA.nex as NEXUS

Data from:
Hayasaka, K., T. Gojobori, and S. Horai. 1988. Molecular phylogeny
and evolution of primate mitochondrial DNA. Mol. Biol. Evol.
5:626-644.

… Skipping taxa block …
… Skipping characters block …
Error reading NEXUS file: expected cost
Error reading primate-mtDNA.nex
g = coding; usertype 2_1 = 4 [weights transversions 2 times transitions

OK, fine. But what’s PISH? No clue. I edited the file and removed the assumptions block, as TNT seems buggy to read square-bracket enclosed comments on PAUP blocks. I tried again and it worked but skipped both the characters and taxa blocks. Time to use one TNT’s example files. So I chose the more evident example.tnt. I repeated the above command now using the tnt file as input. Everything works fine, data is read, stored and ready to go. Basically you have a number of commands that you can use to obtain your trees. I used only mult which does multiple random addition sequences plus TBR (RAS+TBR) to find the most parsimonious tree(s). The program is fast for a morphology matrix of 84 characters and 112 taxa. In less than a second the tree is calculated.

But there is a major problem visualizing the tree(s): the so called PISH (Phylogenetic Inference SHell) does not allow scroll. So after obtaining the tree and using the ne command to visualize it you are left with a “tail” of the tree only. After some fidgeting I found a way to export trees to NEXUS format and could check the trees obtained. By using the command export with a dash and filenames as parameters.

tnt*>export -

Conclusions

TNT seems to be a powerful software and should be able to please anyone intending to employ parsimony as their phylogenetic method of choice. It is fast and, according to the documentation, is loaded.

On the other hand it is still (and probably will be for some time) limited and the CLI version for Linux is not the poster version of a user friendly application. There are many drawbacks and it is very difficult to learn the program without a deep study of the manual. Also the documentation is more focused to the GUI Windows version and the commands that can be used in Linux/OS X are not clear at first. In my opinion the worst feature of PISH is the lack of scroll (if there is scroll I wasn’t able to turn it on) especially when the tree visualization is selected.

Overall is TNT is a fair application, geared to its already known audience. Even being free, I don’t see it occupying the place of PAUP, Phylip and MrBayes in the phyloege. Not without major changes.

Update: another review, of the GUI version is available here.

TNT is free!!! Long live TNT.

TNT stands for Tree analysis using New Technology. First published in 1999 by Nixon and Goloboff on Cladistics (where else?) it gives you the possibility doing phylogenetic analysis using parsimony and … No, that’s it. Only parsimony. But there is a bright side: you can actually run it on Linux and Mac, albeit without a user interface. That’s a major step for a program that when first released had an icon depicting an apple with the forbidden sign around it.

I am planning to test it and post a review soon.

Yep, Santa dropped by before the busy Holiday Season to help with the 17th edition of the Bio::Blogs digest. Welcome and enjoy!

New blogs

This last month we welcomed one interesting blog that will have a lot to add to the community. Alexei Drummond is showing us his bag of goods at the Computational Biology and Evolution, self described as “a heady mix of computational science, evolutionary biology and other things that matter”.

Also Pedro suggested a good look at Thirst for Science, a blog that is surely “drinking the koolaid” and has opinions on a myriad of scientific subjects, from RNAs to population genetics.

Not camera shy

Deepak, from bbgm is definitely not camera shy, and he pitches his opinions on business and scientific intelligence.

I believe there will be a time when life science companies will begin looking at biological, chemical, etc information as a core enterprise asset, rather than something for the research crowd

Watch the videocast too.

Just rambling

Publicly, Pedro rambles this month about the possibilities of predicting functional association by mRNA localization

As expected, more specific mRNA localization terms (localizations 30) are more informative for prediction of functional association since fewer terms are required to obtain the same or higher likelihood of interaction.

And he also rambles about the hot hyped topic of personalized medicine. And he is concerned

that all the attention the genomics side of personalized medicine will distort the relative importance of nature versus nurture.

And he is not the only one posting about it.

Frilas

Frilas is the way freelancers are known in Brazilian newspapers, and Pavel, from Freelancing Science, blogs about the possibility of using Virtual Machines to distribute software.

So creating a virtual appliance for, let’s say, BLAST, would mean installing a fresh copy of our favourite linux under VMWare Server with nothing more than necessary libraries, copy of BLAST executables and possibly a web interface. Voilla. Virtual appliance for BLAST, anybody?

Say wha?

Duncan is still worried about the Semantic Web. Nuf said.

Is there a new Omic?

Keith Robinson, from Omics! Omics!, wants to shrink us all. Maybe it would be good to meet the aunts during Xmas. HE wonders about who or how they are shrinking our (mine, yours, theirs) genomes.

The genome is finished — but not so finished. Bits and pieces are still getting polished up, and while they are generally dull and monotonous a gene or two might still hide there

Already tomorrow

Yep, it is already tomorrow there (no matter what day is here!) and after posting about his perl variables and the election, Neil, who is always doing something rather desperate, tells us about interesting blogs that cover some stem cell subjects. Check it out.

My sardine

This month I posted the last SciView interview in 2007, you might want to take a look. Dr Roderic Page was interviewed and he just said

My sense is that large projects don’t always scale well. You don’t get 10x more out of a project simply because it has 10x as much money. You get administrative layers, and a lot of politics. If the big project is really lots of little ones, then it may well succeed. Barcoding is really lots of little projects — everybody grabs their favourite beastie and sequences it.

among other things. Take a look.

Happy Holidays

Bio::Blogs #17 is very short. Everybody is busy finishing their papers and the shopping list. It will be back next month, maybe here, maybe somewhere else if you want to host it. On the meantime, please to not give this to any of your good friends.

Happy Holidays!!!!

I have a Google alerts with bioinformatics as its key word and this morning I received an email that points to this article on Medical News today. I know it has a huge amount to PR in the article, but the first phrase is symbolic:

“CLC bio, the world’s leading bioinformatics solution provider, announced the release of CLC Bioinformatics Database.”

So, I need to ask: have you ever used a CLC product or know someone that uses it? Leave a comment if so, because I never met someone that uses their products, and they are still the “world’s leading bioinformatics solution provider”.

I have been using RepeatMasker to mask common repeats of my sequences. I have just upgraded it to the version open-3.1.8 from open-3.1.6 and the speed gain was incredible. The same 10 Mb sequence that was taking more than a day to run in 3.1.6 took a couple of minutes on the newest. Nice!

I need to correct myself: I was using version open-3.1.6, not 7