The concept of aqueous computing involves the use of large numbers of initially identical molecules to serve as memory registers in a fluid environment. Here, we consider a new approach to aqueous computing where modified nucleotides are used to 'write' on double-stranded DNA molecules to establish the logical values of true or false for a set of clauses. We introduce an implementation scenario where binding proteins specific to each modification can be used to selectively isolate DNA fragments with these modified nucleotides. In addition, we present initial results showing successful incorporation and detection of modifications as well as separation of modified molecules using binding proteins. As there are millions of molecules with corresponding binding proteins, this approach has the potential to yield unlimited computing power as compared with other aqueous computing methods.