for a set of structural variants (from MANTA, DELLY or LUMPY), please could you advise what is the minimal number of PR (paired-end reads) or SR (split - reads) that you would recommend for filtering ? thank you,
You can't usefully filter based on a specific number of pairs unless you know the depth (at a minimum), and a lot of additional information.
It would be helpful if you gave more details, such as the organism in question, its ploidy, the genome size, the read length you are using, the total number of reads and bases you are using, the sequencing platform, the pipeline you used for read preprocessing and mapping, your specific commands for the tools you mentioned, what you expect the outcome to be (are these low-allele-frequency tumor samples, somatic variations, germline variations?), your specific experiment...
Basically, I suggest you repost your question with sufficient information to answer it, because probably most people looked at it once and ignored it as having too little information to answer.