I want to identify virulent genes in my laboratory strain genome by using blastp tool. What should be criteria of E value and percentage identity? I shall be grateful if someone can send me some good literature reports regarding this issue. Suggestions for this research project would be highly encouraged.
If the alignment between your query and hit is covering the domain region, then you can consider the hit as a homologue even if your E value and identity is very bad.
It depends on the error rate that your scientific question can tolerate.
See my response to this question for a lay explanation of Evalues. BEWARE, the top answer is wrong.
https://www.researchgate.net/post/What_is_E-value_exactly_means_and_what_is_it_1e-63_representing?_tpcectx=qa_overview_following&_trid=53ed6da9d11b8b30678b459d_
A small E-value (less than 0.01) is a necessary but not sufficient criterion for detecting orthologs. An E-value measures similarity beyond randomness, but two sequences may share homologous regions without being orthologs, i.e. they are built from the same building-blocks (domains etc) without having similar functions. Rules like 50% Identity in a region covering 50% of the Query has been used, but this is very coarse. Identify the domains (http://pfam.xfam.org/) in Your virulence genes, and look for them With HMMER3 instead of blastp.
for homology searches I use Evalue:1E-5; qcov:50% and pident 35%. I hope it covers only the positive homologs.
- Badges
- Science topic
- Similar topics
- Filing