4
$\begingroup$

I am looking for a project or tool that allows programmatic pairwise alignments of proteins but that takes care with transmembrane regions of proteins. TM regions are traditionally too information poor for many tools to align accurately (Wong et al, 2010).

  • TM-aligner doesn't appear to have a program to download, and is only available via webserver. On the otherhand,
  • PRALINE has some optimisation for TM regions, however was released in 2008 and since then there has been a boom in TM knowledge from structures so does not benefit from that.Currently I am using PSI-BLAST. However,
  • PSI-Blast uses SEG to mask low complexity regions. This will most likely mean that TMHs are masked making the alignment of those regions fairly unreliable.

Primarily I am interesting in sequence alignments. But if anyone knows of a TM structural alignment, that could be good to know.

I think generally that counting idea works for an ideal helix, but a vast majority of TMHs break convention. I wouldn't expect to see a pattern based aligner that was that general. I'd be interested if anyone knows of a tool that juggles that though!

If I understand correctly, Matteo was saying that TMHs follow a simple patter (HHP/H,HHP/H...~7 times). This is not the case. There are too many examples of TMHs that do not match this pattern to assume this pattern would accurately align TMHs. As @M__ suggests, the number of TMP structures has not caught up to their soluble counterparts in the database. I've updated the question to explain why PSI-Blast might not be the best option.

$\endgroup$
1
  • 1
    $\begingroup$ Hmmmm I think you need to state an example, regarding the vast majority are not helices (suspect you are using GPI anchors as the definition). All viral TM proteins I know use helices. $\endgroup$
    – M__
    Nov 27, 2019 at 19:52

1 Answer 1

1
$\begingroup$

Answer from @matteo-ferla converted from comment:

I assume you mean sequence (MSA), not structure alignment. Despite being a structural bioinformatician, I would slap a big caveat against structure based aligners. TCoffee:Espresso is probably the best structure-based aligner (browser and download and API), but the results are not as solid as some non-structural approaches such as training Psi-blast and aligning with that matrix. I am not familiar with MSA of TM regions, however, so this group may be an exception.

Given that helices are a repeat of 2 hydrophobic residues and 1.5 either polar (core-wards) or also hydrophobic (depending on how the helices pack), a pattern based aligner ought to do well from an evolutionary perspective. If you really want from a structure perspective and don't want to know about the cases where a loss of a turn at the top is compensated by the bottom, say, then TCoffee is definitely worth a try

Answer from @m converted from comment:

PSI-blast is a useful tool for deep alignment patterns, albeit it has its critics. I agree the vast majority of in vivo structures lack TM proteins, for different reasons

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.