Questions about processing single-end Hi-C reads from Ultima Genomics platform

[yihangs]

Hi,

I am now using HiC reads from ENCODE such as https://www.encodeproject.org/experiments/ENCSR458VJJ/. According to the guidelines, these Hi-C reads are single-end, generated by Ultima Genomics platform, and they are processed by the pipeline here. I am curious about how this pipeline converts one single-end read to a read pair after the bwa alignment. My understanding is, two chimeric alignment results of a read from bwa actually represent two ends, and the original alignment of the whole read is useless. However, I find some reads that have more than two chimeric alignments, so I don't know in those cases which two chimeric reads are the two ends. Besides, I am not sure if the two chimeric reads should always be one forward-aligned and one reverse-aligned.

I'd appreciate any advice from you. Thanks!