Blasting using Blastx (nucleotide sequence as input) versus Blastp (translated necleotide sequence into protien sequence using EMBOSS transeq as input)

Question

[drsbalaji]

Can someone please confirm whether or not we will generate the same output by carrying out the following two processes on the same machine and same version of Blast+ (in this case Blast+ 2.2.28)?

Process 1: Given a nucleotide sequence file in fasta format. Use Blastx 2.2.28 for blasting and generate the tabular format output (format 6).

Process 2: Translate the given nucleotide file in fasta format into a amino acid file using EMBOSS transeq in fasta format. Use Blastp 2.2.28 for blasting the translated fasta file and generate the tabular format output (format 6).

If not, can you please clarify where the difference is?

Thanks.

Balaji

PS: We tried these processes on a sample data set and did not get the same output from these two processes.

Answer 1

[Michael]

You are not expected to get the same output. Remarks:

• which database are you using? This will have an influence on the result.
• default parameters for blastx and blastp are different
• blastx can handle frame-shifts, frame-shifts influence the score
• transeq and blastx might use slightly different genetic code
• there could be additional possible alignments on different reading-frames, which you do not capture with a simple forward translation
• aside: you are not using the latest Blast version

[drsbalaji]

Thanks Michael for your response.

We used same protein database for both the processes.

We believe we gave the same command line options for both Blastx and Blastp. However, we will recheck on this.

Since Blastx uses standard genetic code, we gave standard genetic code for translation while using EMBOSS transeq and we translated in all the six frames (both forward and reverse).

Yes. We use older version of Blast+ for now. We will use the latest version as we move forward. Does it matter?

Is there any difference in e-value and bit score calculations in Blastx and Blastp?

With these inputs, do you feel that we are going wrong somewhere?

Thanks.

Balaji