5.3 years ago by
Elaborating and generalizing Neilfws's response,
"Converting" secondary structure to 3D structure would be possible only if you knew the 3d coordinates of each atom because there are way too many quirks in protein structure for any piece of software to reliable predict possible 3D structure without homology modeling or extensive simulation. Each residue would need for a dozen assumptions to be made and that increases the number of possible structures exponentially with increasing length.
Generalizing Neil's response, "converting" is only possible when you're stepping to a format that has either the same or lesser information content than the source format. The other way is possible if the additional information is not important, in which case you can substitute dummy values (as with FASTA to FASTQ conversion).