how to predict transmembrane protein structure - (Nov/14/2017 )

Hi all,

What is the accurate way to predict structure/orientation of a transmembrane protein, I want to knock in a tag into the transmembrane protein for detection purpose and am having trouble as of the ideal position to knock in the peptide tag gene.

Thx.

-Thomson-

Check the tools at ExPASy: https://www.expasy.org/resources/search/querytext:transmembrane

-bob1-

bob1 on Tue Nov 14 22:27:00 2017 said:

Check the tools at ExPASy: https://www.expasy.org/resources/search/querytext:transmembrane

Thanks bob1. How accurate was it to predict using computational tool? Is there anything we shall take into account beside amino acid sequence? Eg signal peptide etc?

-Thomson-

As accurate as any of these sorts of tools are. Basically they look for regions of hydrophobic residues flanked by regions of hydrophilic residues, often in groups of 3 or more and with a certain number of residues spacing the phobic/philic parts ( I can't remember how many).

I think there are some structural considerations, something along the lines of hydrophobic alpha helices separated by hydrophilic loops.

-bob1-

Hi, am wondering if anyone has the experience in reading DAS-TMfilter prediction server. I do not know how to read as it does not hv any explanation beside producing results.

=== Result of the prediction === >

# TMH: 10 Q: trusted

@ 46 8.497 core: 36 .. 56 6.270e-11

@ 103 6.179 core: 92 .. 111 2.243e-07

@ 133 4.706 core: 126 .. 141 4.064e-05

@ 166 5.611 core: 157 .. 175 1.663e-06

@ 204 5.554 core: 195 .. 219 2.038e-06

@ 242 5.371 core: 233 .. 249 3.889e-06

@ 271 3.474 core: 266 .. 277 3.147e-03

@ 322 5.424 core: 313 .. 330 3.221e-06

@ 397 6.361 core: 388 .. 408 1.177e-07

@ 443 6.266 core: 432 .. 452 1.649e-07

For eg this, i think it means amino acid seq 36 to 56, 92 to 111, 126 to 141 and so on are the transmembrane regions of the protein? The rest of the numbers indicates what?

-Thomson-

In general, it is highly recommended to use several tools (at least 2) with (slightly) different approaches fro the prediction of this kind of 'generic' features esp. to reduce false positives.

The automated annotation pipeline of Uniprot, checks first for signal peptides predicted by both SignalP and Phobius. Anything predicted by only one tool is discarded. Then uses TMHMM to predict TM domains and will annotate as such only those that overlap with the prediction of TM given by Phobius

http://www.uniprot.org/help/sam

In addition, you can use TargetP to remove any result that would be likely to be a TM protein of mitochondria or chloroplasts if you are only interested in the TM proteins of the cell membrane

-El Crazy Xabi-

El Crazy Xabi on Mon Nov 20 02:51:27 2017 said:

In general, it is highly recommended to use several tools (at least 2) with (slightly) different approaches fro the prediction of this kind of 'generic' features esp. to reduce false positives.

The automated annotation pipeline of Uniprot, checks first for signal peptides predicted by both SignalP and Phobius. Anything predicted by only one tool is discarded. Then uses TMHMM to predict TM domains and will annotate as such only those that overlap with the prediction of TM given by Phobius

http://www.uniprot.org/help/sam

In addition, you can use TargetP to remove any result that would be likely to be a TM protein of mitochondria or chloroplasts if you are only interested in the TM proteins of the cell membrane

@El Cazy Xabi, Thanks so much for the tips!

As you can see, I'm very new in this and will explore more started from the link provided and do my own study. Incase I still couldn't understand the concept, may i contact u in personal messenger here?

Thanks again.

-Thomson-

Thomson on Tue Nov 21 07:50:02 2017 said:

El Crazy Xabi on Mon Nov 20 02:51:27 2017 said:

In general, it is highly recommended to use several tools (at least 2) with (slightly) different approaches fro the prediction of this kind of 'generic' features esp. to reduce false positives.

The automated annotation pipeline of Uniprot, checks first for signal peptides predicted by both SignalP and Phobius. Anything predicted by only one tool is discarded. Then uses TMHMM to predict TM domains and will annotate as such only those that overlap with the prediction of TM given by Phobius

http://www.uniprot.org/help/sam

In addition, you can use TargetP to remove any result that would be likely to be a TM protein of mitochondria or chloroplasts if you are only interested in the TM proteins of the cell membrane

@El Cazy Xabi, Thanks so much for the tips!

As you can see, I'm very new in this and will explore more started from the link provided and do my own study. Incase I still couldn't understand the concept, may i contact u in personal messenger here?

Thanks again.

Sure

-El Crazy Xabi-