affxparser-package: Package affxparser

Description To get started Setting up the CDF search path Future Work Running examples Technical details Acknowledgments License Author(s) References

Description

The affxparser package provides methods for fast and memory efficient parsing of Affymetrix files [1] using the Affymetrix' Fusion SDK [2,3]. Both traditional ASCII- and binary (XDA)-based files are supported, as well as Affymetrix future binary format "Calvin". The efficiency of the parsing is dependent on whether a specific file is binary or ASCII.

Currently, there are methods for reading chip definition file (CDF) and a cell intensity file (CEL). These files can be read either in full or in part. For example, probe signals from a few probesets can be extracted very quickly from a set of CEL files into a convenient list structure.

To get started

To get started, see:

  1. readCelUnits() - reads one or several Affymetrix CEL file probeset by probeset.

  2. readCel() - reads an Affymetrix CEL file. by probe.

  3. readCdf() - reads an Affymetrix CDF file. by probe.

  4. readCdfUnits() - reads an Affymetrix CDF file unit by unit.

  5. readCdfCellIndices() - Like readCdfUnits(), but returns cell indices only, which is often enough to read CEL files unit by unit.

  6. applyCdfGroups() - Re-arranges a CDF structure.

  7. findCdf() - Locates an Affymetrix CDF file by chip type. This page also describes how to setup default search path for CDF files.

Setting up the CDF search path

Some of the functions in this package search for CDF files automatically by scanning certain directories. To add directories to the default search path, see instructions in findCdf().

Future Work

Other Affymetrix files can be parsed using the Fusion SDK. Given sufficient interest we will implement this, e.g. DAT files (image files).

Running examples

In order to run the examples, data files must exists in the current directory. Otherwise, the example scripts will do nothing. Most of the examples requires a CDF file or a CEL file, or both. Make sure the CDF file is of the same chip type as the CEL file.

Affymetrix provides data sets of different types at http://www.affymetrix.com/support/datasets.affx that can be used. There are both small are very large data sets available.

Technical details

This package implements an interface to the Fusion SDK from Affymetrix.com. This SDK (software development kit) is an open source library used for parsing the various files formats used by the Affymetrix platform.

The intention is to provide interfaces to most if not all file formats which may be parsed using Fusion.

The SDK supports parsing of all the different versions of a specific file format. This means that ASCII, binary as well as the new binary format (codename Calvin) used by Affymetrix is supported through a single API. We also expect any future changes to the file formats to be reflected in the SDK, and subsequently in this package.

However, as the current Fusion SDK does not support compressed files, neither does affxparser. This is in contrast to some of the existing code in affy and relatives (see below for links).

In general we aim to provide functions returning all information in the respective files. Currently it seems that future Affymetrix chip designs may consists of so many features that returning all information will lead to an unnecessary overhead in the case a user only wants access to a subset. We have tried to make this possible.

For older file, certain entries in the files have been removed from newer specifications, and the SDK does not provide utilities for reading these entries. This includes for instance the FEAT column of CDF files.

Currently the package as well as the Fusion SDK is in beta stage. Bugs may be related to either codebase. We are very interested in users being unable to compile/parse files using this library - this includes users with custom chip designs.

In addition, since we aim to return all information stored in the file (and accessible using the Fusion SDK) we would like reports from users being unable to do that.

The efficiency of the underlying code may vary with the version of the file being parsed. For example, we currently report the number of outliers present in a CEL file when reading the header of the file using readCelHeader. In order to obtain this information from text based CEL files (version 2), the entire file needs to be read into memory. With version 3 of the file format, this information is stored in the header.

With the introduction of the Fusion SDK (and the next version of their file formats) Affymetrix has made it possible to use multibyte character sets. This implies that character information may be inaccessible if the compiler used to compile the C++ code does not support multibyte character sets (specifically we require that the R installation has defined the macro SUPPORT_MCBS in the Rconfig.h header file). For example GCC needs to be version 3.4 or greater on Solaris.

In the info subdirectory of the package installation, information regarding changes to the Fusion SDK is stored, e.g.

1
2
3
    pathname <- system.file("info", "changes2fusion.txt", package="affxparser")
    file.show(pathname)
  

Acknowledgments

We would like to thanks Ken Simpson (WEHI, Melbourne) and Seth Falcon (FHCRC, Seattle) for feedback and code contributions.

License

The releases of this package is licensed under LGPL version 2.1 or newer. This applies also to the Fusion SDK.

Author(s)

Henrik Bengtsson [aut], James Bullard [aut], Robert Gentleman [ctb], Kasper Daniel Hansen [aut, cre], Martin Morgan [ctb]

References

[1] Affymetrix Inc, Affymetrix GCOS 1.x compatible file formats, April, 2006. http://www.affymetrix.com/support/developer/
[2] Affymetrix Inc, Fusion Software Developers Kit (SDK), 2006. http://www.affymetrix.com/support/developer/fusion/
[3] Henrik Bengtsson, unofficial archive of Affymetrix Fusion Software Developers Kit (SDK), https://github.com/HenrikBengtsson/Affx-Fusion-SDK


affxparser documentation built on Nov. 1, 2018, 2:25 a.m.