Dear David,

I am not sure whether I understand your question correctly, but perhaps you are familiar with:

Heliquest (containts a lipid binding prediction tool)

MCPep (user-friendly monte carlo program to check whether a segment is surface seeking or transmembrane)

See link or added file for an example of an Eisenberg plot approach which might help:

Article New User-Friendly Approach to Obtain an Eisenberg Plot and I...

Kind regards.

Thank you Rob that was useful, but I think what would settle my mind more is knowing how to use the output data(files) from programs like TOPCONS or TMHMM2 to determine if a protein sequence is a transmembrane protein or at least contains a TM domain. Should I use a consensus or a threshold metric for confidence scores for the predicted TM segments?

Like if I give one of these programs a non TM membrane protein will they output predictions where the secondary structure segments will have low confidence? Or will they output a line somewhere in the file saying not a TM protein/domain?

See enclosed what you get with TOPCONS when you look at Pam16 for example. This is (presumably) a peripheral bound protein. TOPCONS will tell you whether a region is a true TM helix or not, so yes they tell you if it is not a TM protein/domain.

In other words if the program tells you it is, you can be pretty sure it is a membrane protein you are looking at (with my earlier suggestion you can double check).

Hope this is of any help, good luck with your work David.

For those who are interested in how this looks see (or enclosed file):

Data keller3biotech2015suppl.docx

Thank you Rob Keller what I will do now is use TOPCONS2 and if predicted transmembrane helices are longer than 8AAs I will use that as classification of predicted transmembrane domain region.