Fuzzier Logic

Parsing Thermo Finnigan RAW files

Simon — Fri, 11 Jun 2010 09:34:38 +0000

In a rare move, I’m going to largely copy across a post from my work blog, because I hope it contains useful information. For background, I’m trying to write a simple python script that extracts particular metadata from a .RAW file, produced by a Thermo Finnigan mass spectrometer. Tools that exist for parsing these files require access to proprietary XCalibur libraries, which I do not have.

Thermo provided a link to MSFileReader, a ‘freeware’ COM object that should allow interaction with RAW files without an XCalibur installation. They also sent a PDF guide to the COM object. Although this will allow XCalibur to be avoided, the work is still Windows-bound.

Python and COM objects

Python can talk to COM objects, through the win32com.client package. As a test, I installed Python and MSFileReader and the pywin32 libs on my netbook (which is a Windows 7 machine). Can import the required Python module, but need to extent the PATH somewhat:

>>> sys.path.append('C:\\Python26\\Lib\\site-packages\\win32')
>>> sys.path.append('C:\\Python26\\Lib\\site-packages\\win32\\lib')
>>> from win32com.client import Dispatch
>>> x = Dispatch("NAME")

The key thing here is “NAME”:

The provided PDF gives C snippets for each method available in the COM object. This only provides one clue as to the possible name of the COM object

// example for Open
TCHAR* szPathName[] = _T(“c:\\xcalibur\\examples\\data\\steroids15.raw”);
long nRet = XRawfileCtrl.Open( szPathName );
if( nRet != 0 ) {
::MessageBox( NULL, _T(“Error opening file”), _T(“Error”), MB_OK );
…
}

XRawfileCtrl is used to call the Open() method. However, this and MSFileReader as “NAME” both fail (Invalid class string).

Found ‘multiplierz‘ which seems to use MSFileReader to create mzAPI – which focusses on access to the actual data, rather than the metadata. The code gives some good clues as to how to use the COM object. [doi:10.1186/1471-2105-10-364]

MSFileReader.XRawfile is used as “NAME” in this code.

So:

>>> sys.path.append('C:\\Python26\\Lib\\site-packages\\win32')
>>> sys.path.append('C:\\Python26\\Lib\\site-packages\\win32\\lib')
>>> from win32com.client import Dispatch
>>> x = Dispatch("MSFileReader.XRawfile")
>>> x.Open("C:\\Users\\path\\to\\file\\msfile.RAW")
>>>

To be continued…

Telomerase – make your skin immortal!

Simon — Wed, 31 Mar 2010 09:25:21 +0000

I know that the beauty industry has made a habit of twisting science somewhat for it’s own ends (see this and this for instance), but this one takes the biscuit.
The wife spotted a piece in Harper’s Bazaar while she was in the hairdressers yesterday, about an amazing new beauty treatment (the article itself is hard to link to, but it’s number 3 in the list of “9 Skin Secrets for Spring“). Injections of telomerase for $1,500 a pop. Apparently it ‘stimulates resting stem cells’. Obviously the Harper’s piece has guff about it being Nobel-prize winning technology.

Telomerase is an enzyme that amplifies DNA repeats at the ends of chromosomes, without this activity, the telomeres would get progressively shorter until the “Hayflick limit” is reached and the cell will stop dividing, or undergo programmed cell death (there’s a reasonable review of the role of telomerase here: http://www.jco.ascopubs.org/cgi/content/full/18/13/2626).

Now I’m no expert, but as far as I know, telomerase is turned off in normal somatic cells, and telomerase activity has been associated with up to 90% of cancers (even its Wikipedia entry will tell me this much, a rather old paper with some concrete figures can be found here: http://dx.doi.org/10.1016/S0959-8049(97)00062-2). I’m not suggesting for a second that injecting telomerase will give you cancer (the overwhelming probability is it will do nothing at all), but this seems to be an amazing example of abusing science in the name of ‘beauty’.

Posted via email from Simon’s posterous

Stack Overflow, BioStar and – shock, horror – some code

Simon — Fri, 12 Mar 2010 10:12:30 +0000

Stack Overflow reached critical mass a few months ago now. The site gets upwards of 6 million unique visitors a month. The chances are, if you write code, you know it exists, and you’ve received some sort of help there, whether directly or by proxy. In its wake, Stack Overflow has spawned Stack Exchange, a question and answer platform that anyone can buy into, and set up a site. So there’s now Stack Overflow type sites for any number of topics.

Amongst these sites, there have been a couple of attempts to set up science Stack Exchanges (http://asksci.com/ & http://science.stackexchange.com/), but to my mind, science as a whole is not specific enough, even biology is probably too broad an area for Stack Exchange to work as a platform. As a result questions are too hand-wavy, and communities have not really seemed to build. The key to Stack Overflow’s success is that it has very tightly defined boundaries, only questions about programming are accepted, anything else is removed for being off-topic. The site’s creators even set up more sites, Super User, and Server Fault, to keep Stack Overflow on topic.

It seems, then, that bioinformatics is the perfect use case for Stack Exchange. A more narrow domain than science or biology, with an already web savvy community ready to coalesce around a useful focal point. Until recently, however, no one had made the site. Then a couple of weeks ago, http://biostar.stackexchange.com/ started to get some attention on Twitter and FriendFeed. It’s early days, but the site has made a good start, some interesting questions, with some good, intelligent, answers. My main concerns would be:

Not enough users, no critical mass achieved.
- The site seems to be gaining some traction, and getting more active by the day.
Not enough questions, no reason for users to keep coming back.
- This does remain an area for concern, but is also starting to pick up a little.
No financial backing, the site may disappear after the test period comes to an end.
- Stack Exchange is not a cheap platform, and a site like this will need funding to continue. However, the administrator, Istvan Albert, has insisted on the Google Group, set up for ‘meta’ discussion surrounding the site, that the site is funded for a year at least.

Finally, the real motivation for this post… A question on BioStar made me revisit some semi-abandoned code in order to post an answer, and I thought it was quite a nice snippet. About 15 lines of Python that utilises the UniProt ID Mapping service to automate protein ID conversion. I’ve stuck the code into a gist, and I thought I’d stick it up here too.

Posted via email from Simon’s posterous

Impact factors, Colossus and the Wakefield retraction.

Simon — Tue, 02 Feb 2010 21:19:28 +0000

(Graphs from The Independent (London), 21 June 2008)

Today was one of those days where lots of interesting stuff turns up. On the BBC, there was 2 very good pieces about the flaws in the scientific process, specifically closed peer review and impact factors.

I also notice that the BBC are running a daily piece about the history of computing in the UK this week, parts one and two have already been published. Today’s article about Colossus is especially good.

Also, after last week’s excellent, and damning, judgement from the GMC -

- regarding Andrew Wakefield’s reprehensible behaviour in his research into the ‘link’ between MMR and autism, today The Lancet finally pulled the paper in which his findings were published 12 years ago. Wakefield et al (1998) (doi:10.1016/S0140-6736(97)11096-0) has now been retracted from the public record after the Lancet concluded that the claims made by the researchers were ‘false’ (http://www.thelancet.com/journals/lancet/article/PIIS0140-6736%2810%2960175-7/fulltext – apologies for paywall).

Posted via email from Simon’s posterous

Wildlife photographer stripped of award

Simon — Thu, 21 Jan 2010 09:21:52 +0000

via news.bbc.co.uk

The winner of the Wildlife Photographer of the Year award has been disqualified after judges ruled that the featured wolf was probably a “model”.

It is tough to imagine how you would go about getting this photo of a wild animal. The story given by the photographer on winning the prize was reasonably convincing, however.

Posted via web from Simon’s posterous

Summary of the BBSRC Systems Biology Grant Holders’ Workshop 2010

Simon — Thu, 14 Jan 2010 17:42:20 +0000

You could be forgiven for thinking that this meeting was doomed from the outset, heavy snow across the UK making travel very difficult, even for those already in Edinburgh. Those of us in Newcastle just about managed to arrive on time, but there were plenty that didn’t. Despite this, a high proportion of posters still made it, and all but one talk was delivered, though admittedly not by the intended speaker in all cases. Indeed, the first day was like an extended and cruel session of Powerpoint karaoke at one point, with people delivering talks having never seen the slides.

What emerged was a healthy picture of a vibrant systems biology community in the UK, although one tinged by an uncertainty lurking in the not too distant future. Talks from the Centres for Integrated Systems Biology (CISB) directors on the first afternoon gave a flavour of the kind of research being done, and more detailed talks over the following day and a half filled in the picture.

Talks of note (to me at least)
I really enjoyed the keynote, from Peter Smith, of the University of Aberdeen. He is certainly doing systems biology, but not in the way most people first think of it. The ‘system’ he is concerned with is not a cell, an organ, or even a whole organism – it’s a whole ecosystem, or sometimes maybe even larger than that. Systems biology on a continental scale. He painted a very interesting picture of how taking biological factors into account in warming models can make a real difference to predictions. There was also an interesting story of using ecological models to predict bioenergy crop yield, and how this can feed into efforts to introduce certain traits into plants, in order to increase their yields in desirable parts of the globe.

The MCISB talk was great, a really good example of how modelling and ‘omics technologies can dovetail really nicely in a properly integrated project. They’re studying metabolism in yeast, looking at metabolite flux and the reactions that best control it. Flux Balance Analysis, metabolomics, proteomics and ‘traditional’ bioinformatics all combine nicely to illuminate different parts of the process, with data being deposited in many public repositories along the way.

The talk from the PRESTA project was also very impressive. Using lots of microarrays in interesting ways to investigate the stress response of Arabadopsis . Examining at more than just the changing expression of genes in isolation, but looking for patterns of genes that change together in interesting ways, in opposition, or a subset that react to the change in another subset, etc.

The BBSRC systems biology initiative is coming to an end. It was intended to kickstart large-scale systems biology research in the UK through a number of large grant awards, including the CISBs and the Systems Approach to Biological Research (SABR) projects. It was clear from some of the discussions at the workshop that some people are concerned that the winding up of these big grants, coupled with the looming shadow of considerable cuts, means that there is little future for the posse of systems biologists who have been produced over the last few years. ‘Too generalist’ was the major complaint, suggesting that systems biologists are jacks of all trades but masters of none. I would disagree. Systems biologists are masters of systems biology (I know this sounds rather obvious, but bear with me). Traditional biologists tend to be scared by maths, and modelling, and don’t really know where it fits in with what they do. Mathematicians and computing scientists tend to not know anything about biology. There is a large cohort here that knows a great deal of both. Mathematicians who are biology literate, and biologists who aren’t afraid of mathematics and computers. These are skill sets to be prized, not shunned. Systems biology is a discipline in its own right now, and I would be shocked to discover that these researchers do not have a life beyond their Centres.

P.S. I took a poster along, as a part of my role on the Ondex project, it’s on SlideShare, and I’ve embedded it here.

An Ondex dataset for in silico drug discovery

View more presentations from Simon Cockell.

Posted via email from Simon’s posterous

5 Best Data Visualization Projects of the Year – 2009 | FlowingData

Simon — Wed, 16 Dec 2009 09:24:06 +0000

via flowingdata.com

The top visualization project of the year, according to FlowingData, is a project by Ben Fry, which shows changes to the theory of evolution over time. The project takes advantage of the publication of the full works of Darwin online to trace changes in the text of On the Origin of Species over time.
A deserved winner, and very apt, considering the batch of anniversaries that have gone by this year.

Posted via web from Simon’s posterous

A geeky selection of birthday books

Simon — Mon, 14 Dec 2009 10:35:39 +0000

“Beautiful Data”, “Coders at Work” and “xkcd volume 0″

Posted via web from Simon’s posterous

Google Acquires AppJet – are there any live, functional alternatives to Etherpad?

Simon — Fri, 04 Dec 2009 22:40:50 +0000

We are happy to announce that AppJet Inc. has been acquired by Google. The EtherPad team will continue its work on realtime collaboration by joining the Google Wave team.

[...]

The EtherPad site will stay online through March 2010 with some restrictions.

[...]

No new free public pads may be created. Your pads will no longer be accessible after March 31, 2010, at which time your pads and any associated personally identifiable information will be deleted.

[...]

via etherpad.com

Etherpad was a nice little tool, very effective at what it offered, I’m sure the guys who developed it will bring a lot to the Wave party. But seriously, Wave is nowhere near functional yet, it’s confusing, and glacially slow. So is there a decent alternative to Etherpad that is usable – right now?

Posted via web from Simon’s posterous

A test post

Simon — Tue, 01 Dec 2009 22:39:02 +0000

I’ve been writing my blog for about 15 months now. In that time I’ve created a grand total of 33 posts. It’s mostly original content, and I don’t have much time for creating original content. I’ve been trying to think about ways of increasing my posting frequency, and there’s basically 2 things I can do – 1) find more time to generate content, or 2) post stuff that takes less time to generate.

Option 1 is basically an impossibility, given our current understanding of the laws of physics. That leaves option 2.

So I’m going to try cross-posting the content of my Posterous blog to my main site. I generate more posts at Posterous, mainly because it is recycled content, and the barrier to creating posts is so low, I just send an email, or use the bookmarklet that grabs content right from webpages.

Step 1 – stop auto-posting Posterous content to Friendfeed. I’ll let my blog propagate there, and not double-post.

Step 2 – enable XMLRPC posting on my WordPress site.

Step 3 – Turn on WordPress cross-posting on Posterous.

Step 4 – Write this post, and see if it propagates properly.

Posted via web from Simon’s posterous