Secondary structure prediction correctness meassure

Question

[ldpubsec]

Dear everyone, various secondary structure predictors from amino acid sequences produce different results, that's ok, but is there any way (a meassure, or something) that can hint me which prediction is likely to be better than the others? Thank you :)

[jiwpark00]

PSIPRED is the one that I used most. Is that the one you are using? I usually try to combine that result with conserved motifs (i.e. secondary helixes or sheets) should be more conserved.

[ldpubsec]

I'm using PSIRED, GOR and Spider2, then, is the reliability of the prediction rather a question of personal experiences?

Answer 1

[Erik Wright]

Typically there is an associated probability for the prediction at each position. For example, the GOR method provides probabilities for H, E, or C states. These can be calculated in R using the DECIPHER package:

library(DECIPHER)
aa <- AAStringSet("MALKNFNPITPSLRELVQVDKTNLWKGRP")
PredictHEC(aa, type="probabilities")

I hope that helps!

[786]

Hi I want to predict the secondary structure of amino acids of protein by retrieving it from PDB in R. How can I use PredictHEC? What could be the raw code for this purpose?