One of the advantages of ANNOVAR is the relatively fast speed for annotation. The annotation is mostly based on pre-compiled annotation databases, without the need to generate new multiple sequence alignments or the need to interrogate remote SQL databases. To further demonstrate the performance and efficiency of ANNOVAR, we tested it on several additional data sets (Table 2), using a modern 64-bit Linux computer equipped with 3GHz Intel Xeon CPU and 8GB memory. First, we ran ANNOVAR on ∼1 million SNP markers on the Affymetrix Genome-Wide Human SNP 6.0 arrays, and compared the annotations with those provided by Affymetrix (version na30). We identified 271 SNPs that were annotated as exonic SNPs by ANNOVAR but not by Affymetrix. We manually confirmed that these SNPs are indeed exonic based on the latest UCSC annotation database information. This experiment illustrated the importance of annotating variants on-the-fly rather than relying on pre-calculated annotations that can be easily outdated. Table 2.Benchmark results for gene-based annotation on a computer with 3GHz Intel Xeon CPUGenomeData setNo. of variantsTimingNo. of exonic variantsExonic fraction (%)HumanAffymetrix 6.0 SNP array930 0061
