Not clear how method parameters affect performance
Fabien Campagne, Weill Medical College of Cornell University
25 January 2011
An interesting article because DNA methylation studies are now more often performed with sequencing-based method. Efficient approaches and tools are needed to map these datasets to reference genomes. I found the benchmark description to be a weak point of this manuscript. The benchmark results and methods that are presented do not really discuss or measure how the performance (measured as number of reads mapped per second per CPU core) is affected by changing the parameters of the approach (e.g., seed length) or data characteristics (e.g., read length, total reference sequence length). I think these parameters are likely to impact overall performance so it would have been nice to see this discussed. It would be nice if the authors could comment about the impact of these parameters on performance (perhaps on their web-site?)
Not clear how method parameters affect performance
25 January 2011
An interesting article because DNA methylation studies are now more often performed with sequencing-based method. Efficient approaches and tools are needed to map these datasets to reference genomes. I found the benchmark description to be a weak point of this manuscript. The benchmark results and methods that are presented do not really discuss or measure how the performance (measured as number of reads mapped per second per CPU core) is affected by changing the parameters of the approach (e.g., seed length) or data characteristics (e.g., read length, total reference sequence length). I think these parameters are likely to impact overall performance so it would have been nice to see this discussed. It would be nice if the authors could comment about the impact of these parameters on performance (perhaps on their web-site?)
Competing interests
None declared