I used bwa to align a short read file to a reference sequence. I ran it several times, and I got different results for the runs on the same input dataset. The fist run, it generated a ~280,000 lines "gzipped file". I checked the file on the server, and it was actually a sam file, not gzipped. The second run, it generated ~90,000 lines sam file. The third run, it generated ~130,000 lines sam file. The fourth run, it generated ~220,000 lines "gzipped file" Does anyone meet the similar problem? Thank you very much. Cai