I am an extreme novice and doing my first bioinformatics project looking at in silico peptide vaccine design and for my final section, I needed to search around 250 peptide sequences (all <20aa) through BLASTP on the NCBI NIH web version to identify the specificity of the peptides chosen. I have searched them through with the automatically adjusted parameters for short input sequences, however I am now struggling with the best way to interpret these results as it obviously provides a variety of results ranging from low to high E-values.
I also need to search them against homo sapiens, to ensure the peptides aren't also present in humans, but if I automatically adjust the search, I do not know how to interpret these, as I only want non-host homologous peptides and the E-values are mostly very high but it still comes out with the results
Can anyone provide any advice?
Please do not post similar content in multiple threads: BLASTP short sequences (<20aa) - interpreting results You can request a thread bump.
Since we can't see your results not we know what your input data is it is difficult to provide advice.
You know what you want so you are going to need to decide how to proceed. You can easily limit your search against human by using the taxID (9606) when you do your search. So compare the two searches to see what alignments you can eliminate, if they show up as aligning to human proteins.
Thank you for your response, I know to limit the search against human proteome however it is the interpretation I am confused about.
Is there a threshold that is recommended? At what point is it likely a random event?
For example, I am running the peptide 'ISDLTGVTVDHY'. With the automatically adjusted parameters of threshold 20000 and word size 2, I obtain 100 results. The first result is E-value 16, max score 23.5, query cover 50%, per ident 100%. The last result is E-value 281, max score 20.6, query cover 41%, per ident 100%. Would I remove this as a host-homologous peptide, since results appeared, despite having a relatively high E-value? Or should I be using a different/ lower threshold?
Apologies, I hope this makes sense as, again, I am a complete novice and have no prior experience, so I may be going on an incorrect tangent...
I don't know if there is a simple answer. It will depend your goal is and perhaps as subject matter expert you are the best judge for the actual selection. E-value stats are influenced by size of database (see https://www.nlm.nih.gov/ncbi/workshops/2023-08_BLAST_evol/e_value.html ) so you probably can't simply depend on that criteria. Your blast results may not be directly correlatable with the antigenicity or utility of the peptide as a candidate for your search. This is a very specific query and someone who has experience in the vaccine field may be able to answer.