Prompted by the increase in feature resolution and the availability of multipack formats from array vendors, scientists at the French National Institute for Health and Medical Research (INSERM) recently reviewed existing oligonucleotide probe-design tools to provide insight for the microarray community into which design tools might work best for specific applications.
In April, researchers at INSERM's École Normale Supérieure in Paris published a paper in Nucleic Acids Research describing their comparative evaluation of the available oligo-design tools based on "ease of installation, user-friendly access, the number of parameters and settings available."
As part of the review, they also submitted two test cases to a selection of design programs and used a set of tests for the in silico benchmark of the oligo sets obtained from each type of software.
Specifically, the authors looked at 15 different publicly available design programs, including OligoWhiz, a microarray probe design resource hosted by the Technical University of Denmark; the ArrayOligoSelector, hosted by the University of California, San Francisco; OligoTiler, a resource for large-scale tiling array design hosted by Yale University; and the European Bioinformatics Institute's ArrayDesign tool, also intended for tiling array creation.
According to the paper, the selection of design software is dependent on the "goal of the scientist, depending on factors such as the organism used, the number of probes required, and their localization on the target sequence." The authors present their work as a guide to choosing the "most relevant" design software for a particular research project.
"The evolution of next-generation sequencing and the development of multiplexing high-resolution arrays have opened the way to more flexible custom microarray applications on so-called small-genome organisms," Stéphane Le Crom, an associate professor of molecular genetics development and co-author of the Nucleic Acids Research paper, told BioArray News last week.
Le Crom, who manages transcriptome microarray platforms at École Normale Supérieure, noted that Agilent Technologies and Roche NimbleGen have recently rolled out higher-density chips in multipack formats that have lowered the price of arrays. Agilent, for instance, this year launched its million-feature SurePrint G3 platform for comparative genomic hybridization and copy number variation analysis with the capability to produce one, two, four, or eight arrays on a 1x3 inch glass slide (see BAN 3/3/2009). Roche NimbleGen, meantime, sells 2.1-million probe arrays, which are also available in 3 x 720K and 12 x 135K formats.
"Agilent and NimbleGen are now offering new microarray formats and lowering the hybridization price to around €100 ($140), but to access these new applications, like tiling arrays for CGH, exon arrays, or microRNAs, the probe design step has still to be done locally in the laboratory, and this step is not trivial," said Le Crom.
According to Le Crom, oligo design represents a "bottleneck" because designers "need to cope with a lot of parameters for which the real importance is hard to evaluate" as well as a lack of software that helps the user to ensure of design quality. Designs are often validated after an experiment.
"For example, if you design 10 probes targeting one gene you can get seven hybridization results that are comparable and the last three very different," Le Crom said. "A biological explanation could be that in the sample you are looking at there is alternative splicing or different isoforms of your RNA. But another explanation could be that the design of the three probes was not correct."
To navigate these issues, Le Crom and his colleagues focused their comparison on oligonucleotide designs longer than 40-mers for organisms for which the complete sequence is known. To validate the probes, they used two test cases: tiling arrays and custom gene set design. As part of the validation process, the researchers first compared for each of the selected oligo design tools the parameters used to ensure specificity and sensitivity of the designed probes.
As a follow-up, they looked at two real cases where they designed probes with all the available software tools and compared the sensitivity and specificity of the resulting oligonucleotides.
The results of the study showed that the choice of array design software depends on the informatics support available to a given user, according to Le Crom. For biologists, though, Le Crom and his colleagues recommended OligoWiz as a "good compromise between probe design yield and sensitivity with a great graphical interface and tutorial."
For more IT-savvy array users, the authors said that ArrayOligoSelector was able to "achieve great results in term of sensitivity and specificity with the ability to design probes in almost every case." Finally, for tiling arrays, the authors argued that the "two best and available solutions seem to be OligoTiler and ArrayDesign."