CNV detection in cancer sample can be done based on either mapping distance of paired-end reads or read depth. However, The mapping distance strategy is not able to detect CNV in case of aneuploidy, right? Because aneuploidy produces CNV while the mapping distance of paired-end reads on each chromosome still remain unchanged. Does this mean the read depth strategy is better since it cover both situation? Can someone share some insight?
Read depth is very noisy due to amplification biases, general uneven amplification and random sampling from a library. So it's definitely not the holy grail of CNV calling methods. As often, the most powerful methods combine read depth, mapping distance and split reads to get the most reliable calls.