Maybe a quantitative RT PCR, using primers binding specifically to one essential part of viral DNA. And other primer pairs binding to some host gene that can act as internal control. In case that the copies of the viral DNA will increase with respect to the host DNA (in certain conditions), you can know that the virus is replicating. This will be only for detection, not for isolation. Just the first idea that cross my mind, but I´m sure that virus experts can propose something better. 

Hi Puja,

your viral DNA is most likely non homologous to your host DNA, so it should not be a problem to identify your virus without any worries from human DNA. 

Do you want to test for a specific virus or test a sample for an unknown virus? 

If you want to test for a specific set of virus (ex. HSV), real time PCR with good pair of primers would be the technique of choice. You have to extract the DNA using an appropriate technique (like HighPure® Viral Nucleic Acid Kit) and then run your realtime PCR. If your test if positive, you will know that the virus is there. These techniques are low cost, fast and reliable. You can also check if a multiplex PCR exists for your virus of interest. 

If you want to test a sample for the presence of a virus or identify new viruses, the gold standard would be a next-generation sequencing pipelines. You have to consider many drawbacks and pitfalls before you commit to a certain pipeline like read-lenghts, viral particules enrichment, etc. This recent review, by Datta et al., should provide you with all the information you need to get started!

Best,

Nicolas 

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4534817/