Why do people suggest that translated sequences be used to search for relatives in databanks? 


There are three main purposes for conducting a fast database search (FDS): to find out quickly if a new nucleic acid sequence in hand has been already deposited in the databanks, to find similar non-coding DNA stretches in the databanks (e.g. repetitive or regulatory sequences), and to find proteins homologous to a putative coding open reading frame (ORF). An ORF is one of the three or six (if both strands are considered) possible conversions (reading frames) of a DNA sequence to an aminoacid sequence. If an ORF is over 200 nucleotides long and is not interrupted by stop codons, it might be considered for further analysis. Among other uses that have gained popularity recently are: to validate a predicted coding ORF by finding homologous polypeptides in databanks , and for locating false priming sites for a set of PCR oligonucleotides or probes. 

For the search of specific DNA sequences, it is obvious that a comparison of the raw nucleic acid sequences against GenBank/EMBL is appropriate. 

However, in the case of conducting a search to find distantly related homologous proteins to a putative coding ORF, it is essential to translate the ORF to the corresponding aminoacid (aa) sequence in order to increase the sensitivity of the search. In the field of molecular biology, homologous means a similarity derived from common ancestry, which in turn implies a similarity in the amino acid sequnce or in the folding or function of the protein. DNA sequences are composed of a four-character alphabet. It has been pointed out that when using DNA sequences in a FDS, it is easy to find similar sequences which are not related by ancestry or function, simply due to random events or convergent evolution. I cite Doolittle: 

"DNA is composed of only four kinds of units -A, G, C and T- and even if gaps were not allowed, it would be anticipated that, on the average, 25% of the residues of any two aligned sequences would be identical. In fact, there would be a dispersion around the mean expectation, and a predictable fraction of random cases would be as much as 35% identical. Once we decide to allow gaps in the sequences, then the range of chance similarities between two unrelated sequences can exceed 50%, thereby obscuring any genuine relationships that may exist" 

On the other hand, protein sequences are composed of a 20 aa alphabet determined by 61 degenerate codons. When the DNA sequences are translated into 21 different types of codons (20 aa and a terminator), the information is sharpened up considerably. The 'wrong-frame' information is discarded, and third-base degeneracies are consolidated. All in all, the signal-to-noise ratio is greatly improved for the specific purpose of identifying protein relatives. It is accepted that convergence phenomena in aa sequences are very rare and thus aa similarity almost always means homology. Furthermore, aa sequences may still show a similarity derived from common folding patterns and function of the proteins, even while their coding DNA sequences might have strongly diverged due to other selective pressures existent at the genome level (e.g., G+C pressure, preferential usage of synonymous codons, etc.). Protein evolution is governed by the constraint of maintaining a characteristic fold which enables some function. Thus, it is possible to infer relationships between proteins that last shared a common ancestor 1-2.5 billion years ago by conducting protein searches, doubling the lookback time obtained performing DNA database searches. 

Finally, when validating putative coding frames by finding homologous polypeptides in the databanks, the search should be conducted using aa sequences for the same reasons as given above. It is also useful to carry out the search with a DNA databank translated in the six possible ORFs (e.g., TFastA and tBLASTn programs). It is not rare to find similarities with partial or complete putative polypeptides not described by the authors who deposited the sequences in the databanks.