Need help with AUGUSTUS and AGAT for gene prediction and feature extraction, respectively.

Question

[Vijith]

I have two queries: one is regarding AUGUSTUS, and the second is about extracting sequences from the *.gff file for downstream BLASTx homology-based annotation.

1. I ran AUGUSTUS using the command

    augustus [parameters] --species=SPECIES queryfilename > output.gff

without specifically setting --alternatives-from-sampling to True. Will this affect the completeness of the downstream annotation process?

1. After completing the de novo gene prediction, I want to run BLASTx for homology-based annotation, alongside the evidence-based approach. However, I’m confused about extracting features from the *.gff output. In a previous faulty AUGUSTUS run, I used AGAT's **agat_sp_extract_sequences.pl** to extract sequences with the command.

agat_sp_extract_sequences.pl -g infile.gff -f infile.fasta -t gene

However, after reading more about its suitability for BLASTx, I realized this approach might include introns, UTRs, intergenic regions, etc. Therefore, I’m considering using the alternative command:

agat_sp_extract_sequences.pl -g infile.gff -f infile.fasta --mrna

or

agat_sp_extract_sequences.pl -g infile.gff -f infile.fasta -t cds.

If you want details on what these commands extract, please take a look at this image.

Answer 1

[Juke34]

Since blastx translates the query sequence in all six reading frames to blast against a protein database, the logic would push you to extract only what is supposed to be translated i.e: CDS.

So use the last command e.e:

agat_sp_extract_sequences.pl -g infile.gff -f infile.fasta -t cds

[Vijith]

Juke34, thank you so much for your valuable response. This helps.