Protein prediction from cancer genome is difficult. You may need transcriptome (RNA-Seq) for that. From genome, you can get mutations, insertion, deletions and translocations. You may not need de-novo assembly for that. Alignment to existing genome should be sufficient.

One thing you could do is to infer a list of proteins that are modified in cancer genome due to specific mutations leading to modifications in the CDS of transcripts (premature polyA, frameshifts, etc...)

Dear Satya & NicoBxl,

Thanks for your response.

I think my question is not clear. Please let me know whether this procedure can be followed or not.

Denovo Assembled Cancer Genome -- > ORF Finders -- > Protein prediction.

I dont want to miss any single protein that can be translated from a cancer genome. My interest is sometimes these proteins may not be of human origin.

Please update me waht can be done further.