Entering edit mode
4.7 years ago
Chirag Parsania
★
2.0k
tl;dr; I wonder , is there any known tool / webserver / r package to check quality of MSA?
---- EDIT -----
The idea here is, for a given protein sequence to find out set of orthologues and perform the phylogenetic analysis. To do so, I performed blast and got first set of orthologues. However, to get complete set of orthologues I performed multiple blast with different start seed. At the end I combined all the confident (evalue <= 0.00001, qcov =90%, % identity = 40 %) blast hits. Next, I performed MSA. Now, I want to check whether given MSA is confident and represent set of orthologues for a given query protein.
It is difficult to objectively asses the quality of alignments without a reference. What exactly do you have in mind?
There are several databases with reference alignments that are used to benchmark multiple alignment programs. See here for one example, but there are others.
Hi, Thanks for reply. I updated original question. Kindly check.
Same answer as before: it is difficult to assess alignment's quality without a reference.
It seems to me that rather than assessing quality you are trying to find outliers in your alignment. That can be done.
https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-015-0702-1 https://journals.sagepub.com/doi/full/10.4137/EBO.S40583
Before attempting any of these, I suggest you set a more stringent E-value. What you are using (1e-5) may be OK for detecting paralogs, but I think it is not stringent enough for orthologs. I've seen others use 1e-20 for detecting orthologs, but you could research that part yourself by checking what BLAST E-values are used by OrthoMCL or similar programs.