Integral membrane proteins are proteins that have at least one transmembrane (TM) segment. I performed predictions of TM domains using three tools: ProtScale, ΔG predictor 1.0, TOPCONS.

For example, the results in terms of aminoacid positions for the TM1 domain were as follows:

  • ProtScale: 37-55
  • ΔG predictor: 37-55
  • TOPCONS: 36-56

The limits of the predicted TM domains can be used for drawing the topology of this protein.

For this purpose, we need to have representative limits from these results.

Statistically speaking:

  • which values should be chosen?
  • should we calculate a mean : (37+37+36)/3=36.67 knowing that these values correspond to aminoacid positions and 36.67 doesn't mean anything?

How can I use these results (37, 37, 36) to determine the limit corresponding to the 1st aminoacid of TM1 segment?

  • $\begingroup$ How big is the protein? Normally TMDs are at the 3' end of the protein. $\endgroup$
    – M__
    Jul 18, 2022 at 15:12
  • $\begingroup$ The protein has 426 aminoacids. TM1 is the first TM segment, from the N-terminal domain. $\endgroup$ Jul 20, 2022 at 21:40

1 Answer 1


The variation you are seeing are minor differences in estimates in the transmembrane domain, these are usually modelled by hydrophobicity and are accurate predictions. If you look at the secondary/tertiary structure you might see alpha helices which will 'burrow' into the membrane. The region in this particular question is a small domain however, but the position is 'about right'.

Thus question is whether 1 amino acid at 5' and 3' site is inside or outside the membrane - I think this is overthinking. Edge-effects are common in any estimate and 1 amino isn't bad variation at all. To double check have a look at the Kyte-Doolittle index on the Expasy website (which is one of the first), and TMHMM. Finally, you need to check the crystal structure if available.

What does puzzle me is the apparent absence of a 3' GPI anchor, but I dunno the biological system so ain't for me to comment.


Your Answer

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

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