I am using tblastn to search protein queries against custom eukaryotic genome assemblies that lack protein annotations. One suggestion was to parse the hit coordinates and extract the matching genomic regions. Does the blast toolkit make this possible? Does it also make sense to translate the extracted sequences from the genome just so I can do a reciprocal blast to check whether these sequences are homolog candidates?

Would it be more appropriate to first generate a eukaryotic protein annotation for these genomes and then use blastp instead, or is tblastn the preferred approach in this situation?

Thanks in advance for any input. Not sure if this makes sense.