To compare the speed and trimming results of ITSxpress, we compared the ITS1 and ITS2 sequences from 15 soil samples collected from the rhizosphere of maize in fields with different winter cover crops. ITS1 reads were amplified using the ITS1F/ITS2 primer set (
Gardes & Bruns, 1993 (link);
White
et al., 1990 ). ITS2 reads were amplified using the ITS3/ITS4 primer set (
White
et al., 1990 ). Reads were multiplexed and sequenced on an Illumina Miseq in 2x300bp run mode using version 3.0 chemistry.
Tests of ITSxpress and ITSx performance were run on single compute nodes with 2 x 10 core Intel Xeon Processors (E5-2670 v2 2.50GHz 25MB cache) with hyper-threading enabled, 128GB DDR3 ECC memory and two Intel DC S3500 Series SATA 6.0Gb/s SSDs. For the first test of trimming speed, 5 replicates were run where 15 ITS1 and 15 ITS2 samples were trimmed using ITSxpress and ITSx with 4 logical cores. Trimming was done using ITSxpress with default settings. ITSx was run with multithreading and heuristic filtering turned on and only the fungal database selected. The running times for ITSx and ITSxpress were plotted on a log scale,
Figure 1 . The number of total reads in each sample and reads remaining after clustering at 99.5% identity are shown on a log scale,
Figure 2 .
To compare the performance of ITSxpress and ITSx as computer cores were added, tests were run on the ITS1 and ITS2 sample with the largest numbers of sequences (ITS1: n=100543 16% unique, ITS2: n=145499, 30% unique). The sample was processed 5 times with 1, 4, 8, 16, 30, and 40 virtual compute cores. The mean and standard error were plotted,
Figure 3 . Program settings were the same as in the first test.
The trimming positions from ITSx and ITSxpress were compared for every ITS1 and ITS2 sequence. If a read was not trimmed identically by ITSx and ITSxpress, it was globally aligned and the start and stop positions were compared. Alignment was done using the Biopython Pairwise2 implementation of a global alignment function with the parameters (match score: 2, mismatch penalty: -1, gap opening penalty: -0.5, gap extension penalty: -0.1) (
Cock
et al., 2009 ).
Gardes & Bruns, 1993 (link);
et al., 1990
et al., 1990
Tests of ITSxpress and ITSx performance were run on single compute nodes with 2 x 10 core Intel Xeon Processors (E5-2670 v2 2.50GHz 25MB cache) with hyper-threading enabled, 128GB DDR3 ECC memory and two Intel DC S3500 Series SATA 6.0Gb/s SSDs. For the first test of trimming speed, 5 replicates were run where 15 ITS1 and 15 ITS2 samples were trimmed using ITSxpress and ITSx with 4 logical cores. Trimming was done using ITSxpress with default settings. ITSx was run with multithreading and heuristic filtering turned on and only the fungal database selected. The running times for ITSx and ITSxpress were plotted on a log scale,
To compare the performance of ITSxpress and ITSx as computer cores were added, tests were run on the ITS1 and ITS2 sample with the largest numbers of sequences (ITS1: n=100543 16% unique, ITS2: n=145499, 30% unique). The sample was processed 5 times with 1, 4, 8, 16, 30, and 40 virtual compute cores. The mean and standard error were plotted,
The trimming positions from ITSx and ITSxpress were compared for every ITS1 and ITS2 sequence. If a read was not trimmed identically by ITSx and ITSxpress, it was globally aligned and the start and stop positions were compared. Alignment was done using the Biopython Pairwise2 implementation of a global alignment function with the parameters (match score: 2, mismatch penalty: -1, gap opening penalty: -0.5, gap extension penalty: -0.1) (
et al., 2009
Free full text: Click here
