Birdsuite: Birdseed

• Description
• Binary files
• Library files
• Running Birdseed and Documentation
• Sample data sets

Birdseed Description

Birdseed is SNP genotyping algorithm that runs on the Affymetrix 500K, SNP5.0, and SNP6.0 platforms. Although Affymetrix officially supports Birdseed only for SNP6.0, we and others have found that it has excellent performance on the 500K and SNP5.0 platforms as well.

Birdseed uses a customized Expectation-Maximization (EM) algorithm to fit two-dimensional Gaussians to SNP data, producing genotypes and confidence scores for every individual at every SNP. Because it is a clustering algorithm, it should be run on many samples at a time (typically 50 or more).

Birdseed Binary files

Note: In the summer of 2007 an update was made to Birdseed (Birdseed v2) which further increases clustering robustness by placing stronger reliance on the SNP-specific priors. For maximum robustness we recommend that users should switch to Birdseed v2. At the present time Affymetrix is providing full commercial support for Birdseed v1 and is delivering Birdseed v2 in the "birdseed-dev" section of Affymetrix Power Tools.

The binary files for Birdseed are currently being distributed by Affymetrix. It is supported on the Windows platform, Mac OS X, and Linux (both 32-bit and 64-bit). Birdseed version 2 can be found only in APT (Affymetrix Power Tools) version 1.8.5 or greater. We strongly recommend using APT version 1.8.6 or greater.

Click here to obtain the latest release of APT. Source code is available on the same page.

Birdseed Library files

A number of library files are required to run Birdseed. The required library files are:

1. The chip definition file (.cdf): This is platform specific and is supplied by Affymetrix (available as a free download). It is a binary file that describes the layout of the array.
2. Special SNPs/chromosome X file: The 500K, SNP5.0, and SNP6.0 platforms all have SNPs on chromosome X that must be called differently on males versus females (because of males having a single copy and females having two copies). In addition, SNP6.0 has chromosome Y SNPs and mitochondrial SNPs. The easiest way to handle these so-called "special SNPs" is to supply a file enumerating and classifying all the special SNPs (X, Y, or mitochondrial). For the 500K and SNP5.0 platforms, these files may be downloaded below. A special SNPs list is provided by Affymetrix for the SNP6.0 platform (available as a free download from their website and also below).
3. Gender calling copy-number probe lists (SNP5.0 and SNP6.0 only): Internally APT/Birdseed determines the gender of each sample. This is necessary to handle SNPs on the X chromosome which are single copy in males and two copy in females. The optimal way to determine gender is to examine a ratio of selected copy-number probes on the X versus Y chromosome. These two text files specifying the selected X and Y copy-number probe lists may be downloaded below. (For the 500K platform, a separate method for computing gender is used and no additional files are required.)
4. Models file: This is platform specific and can be downloaded below. It is a text file (tsv format) that contains estimates of where SNP clusters fall based on Hapmap data. Unlike the other library files which are all static, the contents of this file can theoretically improve as more data are acquired. Hence there is a version number tag on each models file; we hope to post improved models files at a later date.

Affymetrix support pages (see under "Library Files" for the relevant files):

• 500K
• SNP5.0
• SNP6.0

Running Birdseed and Documentation

The Affymetrix Power Tools which includes Birdseed comes with extensive HTML documentation on how to run Birdseed. Many options and flags are discussed there that may be useful to more advanced users. Below are standard, basic commands to run Birdseed for the 500K platform, SNP5.0, and SNP6.0. These commands are shown for a typical Linux installation of APT.

Note: You must use as your analysis string "-a birdseed-dev." This will tell the program to run Birdseed version 2. (The analysis string "-a birdseed" will run Birdseed version 1.)

250K (Nsp or Sty) example:

apt-probeset-genotype \
  -a birdseed-dev \
  -o results_dir \
  -c Mapping250K_Sty.cdf \
  --special-snps Mapping250K_Sty.specialSNPs \
  --summaries \
  --read-models-birdseed Mapping250K_Sty.v1.5.birdseed.models \
  --write-models \
  *.CEL

SNP5.0 example:

apt-probeset-genotype \
  -a birdseed-dev \
  -o results_dir \
  -c GenomeWideSNP_5.cdf \
  --special-snps GenomeWideSNP_5.specialSNPs \
  --set-gender-method cn-probe-chrXY-ratio \
  --chrX-probes GenomeWideSNP_5.chrXprobes \
  --chrY-probes GenomeWideSNP_5.chrYprobes \
  --summaries \
  --read-models-birdseed GenomeWideSNP_5.v4.5.birdseed.models \
  --write-models \
  *.CEL

SNP6.0 example:

apt-probeset-genotype \
  -a birdseed-dev \
  -o results_dir \
  -c GenomeWideSNP_6.cdf \
  --special-snps GenomeWideSNP_6.specialSNPs \
  --set-gender-method cn-probe-chrXY-ratio \
  --chrX-probes GenomeWideSNP_6.chrXprobes \
  --chrY-probes GenomeWideSNP_6.chrYprobes \
  --summaries \
  --read-models-birdseed GenomeWideSNP_6.v2.6.birdseed.models \
  --write-models \
  *.CEL

Sample data sets

Soon we will be posting sample CEL files to test your installation and execution of Birdseed.