MSCompBio Forum Rss Feed

New Post: HLA Completion - small typo

delordson — Mon, 14 Sep 2009 05:20:17 GMT

Hi,

There is a small but significant typo in the summary info on the HLA Completion tool. Where it states:

"For example, an allele may be identified as A6801/6802 or simply A02."

This should actually be A28, as A*6801 and A*6802 are both alleles of A28 not A02 (even though A02 is closely related in sequence and reactivity).

Delordson Kallon

HLA Bioinformatics Specialist

National Blood Service, England

New Post: HLA Completion Result Sort Order

delordson — Sat, 12 Sep 2009 08:45:16 GMT

Just heard you and Dot Net Rocks (www.dotnetrocks.com) so I came over here to have a look. The HLA completion tool looks very interesting. I'll play it against some of our data and let you know how it goes.

Can I suggest that if possible, the result s be sorted in decreasing order of probability. I tested it with A*0101 A*0201 B*0801 B*4402 C*0501 C*0701 in a European population and almost dismissed the tool when I saw the results until I took a closer look at the probabilities.

Delordson Kallon

HLA Bioinformatics Specialist

National Blood Service, England

New Post: Epitome Creator: Silverlight for Science

CarlK — Fri, 14 Nov 2008 01:04:33 GMT

Greetings,

We've release a new biology tool (and web toy). It's "Create Epitome". The web version is http://atom.research.microsoft.com/EpitomeCreator/. It can also be run as an off-line web page (via Silverlight) and a command-line app (via .NET).

Along with using it to create HIV vaccines, you can also play with it as a little web toy. If you paste in

Twinkle, twinkle, little star; How I wonder what you are.

You can watch it build up, one step at a time:

Method	AminoAcidLength	numComponents	coverage	Vaccine
Greedy	7	1	0.3	TWINKLE
Greedy	7	1	0.3	TWINKLE
Greedy	10	2	0.4	TWINKLE,HOW
Greedy	13	2	0.5	WHATWINKLE,HOW
Greedy	12	1	0.5	HOWHATWINKLE
Greedy	15	2	0.6	HOWHATWINKLE,YOU
Greedy	18	3	0.7	HOWHATWINKLE,YOU,ARE
Greedy	20	3	0.8	HOWHATWINKLE,YOU,STARE
Greedy	26	4	0.9	HOWHATWINKLE,YOU,STARE,LITTLE
Greedy	32	5	1	HOWHATWINKLE,YOU,STARE,LITTLE,WONDER

Silverlight is great for this kind of tool because

· The port from .NET was easy.

· Processing, which could be long, is on the user's machine not the web server

· No user data need be transferred to our servers

· Runs on Windows, Mac, or (soon) Unix
· Users get rich interaction (including an animated plot) with web-like simplicity.

The .NET command-line version uses the Microsoft Parallel Extensions Library to run muti-core, leading to faster results.

The code is open source and available at:

http://www.codeplex.com/MSCompBio/Wiki/View.aspx?title=Create%20Epitome

Some source code features of possible interest:

· Builds both Silverlight and .NET command-line versions from the same C# code in the same Visual Studio solution.

· Includes installer that installs both versions, including setting up the Silverlight page to run off-line.

· The Silverlight version resizes the Silverlight control on the web page.

· The Silverlight version shows results interactively as they are produced. It allows the run to be paused and continued. It plots the performance of the algorithm.

· The Silverlight control appears in an HTML table and is visible to both IE and Firefox.

· The code uses the HashSet data structure. For Silverlight, where HashSet is not available automatically, a mini version is provided.

- Carl

Carl Kadie
http://research.microsoft.com/~carlk

New Post: HLA Completion: New method

CarlK — Mon, 13 Oct 2008 16:18:33 GMT

Greetings,

We've replaced the HLA Completion tool with a new method based on

Listgarten J, Brumme Z, Kadie C, Xiaojiang G, Walker B, et al. 2008 Statistical Resolution of Ambiguous HLA Typing Data. PLoS Computational Biology 4(2): e1000016 doi:10.1371/journal.pcbi.1000016

(paraphrasing the abstract)
The new method improves upon the Expectation-Maximization (EM)-based approaches currently used within the HLA community. Our improvements are achieved by using a parsimonious parameterization for haplotype distributions and by smoothing the maximum likelihood (ML) solution. These improvements make it possible to scale the refinement to a larger number of alleles and loci in a more computationally efficient and stable manner. We also show how to augment our method in order to incorporate ethnicity information (as HLA allele distributions vary widely according to race/ethnicity as well as geographic area).

- Carl

New Post: Tools relicensed as "Ms-RL"

CarlK — Fri, 10 Oct 2008 22:56:43 GMT

We've relicensed the tools. They are now under the generally more liberal Microsoft Reciprocal License (Ms-RL), which is recognized by the Open Source Initative.
- Carl