Skip to main content

Table 1 Performance of trieFinder versus Bowtie

From: trieFinder: an efficient program for annotating Digital Gene Expression (DGE) tags

  Mean CPU time CPU time SD Mean wallclock time Wallclock time SD Peak virtual memory (MB) Reads aligned (%)
trieFinder 3:00:53 0:00:35 2:24:17 0:00:33 > 4096 84.55
Bowtie 3:46:40 0:03:41 3:47:49 0:03:36 2564 84.51
  1. Performance of trieFinder (commit aca5281183) and Bowtie (version 1.0.0, commit fe7a830e31; 64-bit) when aligning 327,240 unique 20-bp DGE reads from zebrafish, allowing a single mismatch. trieFinder was run using default settings. To match trieFinder as closely as possible, Bowtie-build was used to build three indexes based on the three database files (RefSeq, UniGene, and genomic) with default settings. The UniGene input was modified to remove comment lines inconsistent with the FASTA format when building the Bowtie indexes. Three Bowtie alignments were run, one for each index, using the settings ‘-v 1 -a -y -t --fullref --sam --sam-nohead’. The times represent the mean of the sum of all steps, including database/index creation and the sequence search, for three runs of each program. All runs were performed on a server with 64 GB of RAM and eight 800 MHz Quad-Core AMD Opteron processors. Times are displayed as hours:minutes:seconds. DGE reads are from Liang et al., 2012 [8]. CPU, central processing unit; SD, standard deviation.