Is there any rule of thumb for allocating memory through "bsub" for running DataProcessingPipeline per bam file or per number of reads ?
Sorry, we don't provide recommendations at that level. Perhaps other users may want to share their experience in this matter?
and just to add some other aspect to my questions for other users, what could be an efficient way to retrieve the processing; meaning continuing the pipeline from the point that it got terminated.
I'm a little confused by what you're asking. Queue automatically checkpoints the steps as it runs them, and will restart from the appropriate point by itself (assuming you don't tell it otherwise).
As for the memory, that's controlled in the qscript, not from the bsub command. The actual Queue job that you submit doesn't consume much in the way of resources, I think I give it 4GB of RAM, but that's probably way too much