DNA data or RNA data? NGS sequencing is biased by CG content and many other things which means some regions are always lowly covered. Also consider the mappability of each region, some regions are always can not be mapped. If RNA, you should also think about the expression level of your interested RNA.

DNA data, mate-pair libraries (8kb), I have 3 contigs (2-3 kb) with a low coverage about 60X when the rest of contigs 200X.

"Standard library preparation methods include a PCR enrichment step prior to cluster generation. Biases inherent in PCR amplification result in uneven read coverage and increase the numbers of duplicate fragments present in the library."

http://www.ashg.org/2013meeting/abstracts/fulltext/f130122314.htm