Detect somatic variants in single tumour sample

[dukzilla]

There's a lot of tools that do a tumour/normal pairing exercise to detect somatic mutations but are there tools to detect SNPs/indels in the case where there's just the tumour and no normal to compare to? If I have a tumour sample (and its impossible to separate tumour/normal cells) and prep my DNA from this heterogeneous mass, is there any bioinformatics tools that can identify in any of the detected SNPs are somatic SNPs/indels?

[Yilong Li]

Well you won't be able to know for sure whether a variant is somatic or germ-line, since you don't know what the germline looked like...

Could you do something like: get a confident set of variants -> filter against all the datasets you can possibly find (1000GP, dbSNP, more somewhere???) -> hope you don't get too many remaining variants and validate them using Sanger sequencing or something from both your tumor and matching normal DNA (provided you do have the normal DNA but just didn't sequence it).