I'm mapping 50 bp reads from several ChIP-seq datasets to a short reference (170bp). I want to measure the relative enrichment of this specific consensus in the different IP conditions. It's repetitive DNA so I guess it's ok to use a consensus.
To do so, I used Bowtie2 (with the --end-to-end and --local modes) and Lastz (with the 12of19 seed and no filtering).
The number of reads mapping to my reference in all datasets (about 30 ngs files) was always bigger when I used Lastz. I can understand why bt2 end-to-end VS local give different results. But shouldn't Lastz and bt2 --local give comparable results?
Should I apply filtering to the Lastz mapping?