Hi,

I have a question about dispersion in read numbers. I have experienced more than two times the difference in the read number among samples, which seems unusual. I and understand downsampling is one of the options for the issue. I am wondering:

1. how the dispersion of read numbers affects the downstream analysis if I do not downsample them.
2. which downsampling is better, from FASTQ files or from BAM files after mapping.
3. what are the popular tools for downsampling.

Thank you for your help.