Should I Perform A Local Alignment To Try To Identify Localization Sequences?

Question

[John]

I have a list of 15 genes known to be expressed in the yeast plasma membrane. I'd like to try and identify a common localization tag within these sequences. I think I need to perform a local alignment of the sequences rather than a global alignment. I'm using MUSCLE to do the alignment but I'm a bit overwhelmed by all of the options. I've tried a gap penalty of -1, -10 and -100. -10 seems to provide only 2 residues that align, the -1 shows a global alignment and -100 nothing at all.

Any help appreciated on the best settings to use. Should I keep trying different gap settings or should I change a different parameter, or use a different program?

Many thanks. John

[Mchimich]

0

Try MAFFT software with default parameters. For more detals please have a look on http://mafft.cbrc.jp/alignment/software/algorithms/algorithms.html. cheers

[Whetting]

@John If you are not sure about what each of the options does, I would argue against changing them. (I would also suggest trying to understand them before moving forward). That being said, it depends on how divergent your samples are. The default parameters are likely to be ok. I To go back to the original question: why do you think a local alignment is to be preferred over a global alignment?

[John]

Thanks Whetting. The only relationship between all of these sequences is that they are known to be localized to the plasma membrane. I would not expect a global alignment to be of much use as I don't expect they have a common origin. However, I did think they may have common localization sequences in them, hence I though I'd try a local alignment. I'll investigate the options as you suggest.

Answer 1

[DG]

Aren't there programs for predicting these localization/sorting signals? I'm more familiar with chloroplast and mitochondrial signal prediction than plasma membrane but it seems to me there are a few out there. That would be you're best bet as localization signals of all types tend to share some sequence properties and patterns but aren't usually very conserved even at a local alignment level.

[Whetting]

As was mentioned in a previous post, membrane proteins usually do not really have a "localization signal". I would look for conserved hydrophobic patches. Good transmembrane protein predicting algorithms do exist though