Agadir is a prediction algorithm based on the helix/coil transition theory.
Agadir predicts the helical behaviour of monomeric peptides. It only considers short range interactions.
Conditions such as pH, temperature and ionic strength are used in the calculation. Modifications of the termini are also allowed.
Keep in mind that Agadir is not a program to predict secondary structure of proteins.
1200 peptides were analysed to evaluate the performance of the algorithm. The average deviation of the prediction from experimental values (obtained from CD measurements) is -2 percent of helical content, with a standard deviation of 6 (percent of helical content). i.e., a peptide with 30% helix content would be predicted to have 28%+-6%.
Some side chain-side chain interactions are missing or approximate, due to lack of insuficient data. This is specially true for Pro, His, Cys and Trp.
In long peptides (more than 30 aminoacids), tertiary interactions can develop. Such interactions, even with a half-life below the time scale of NMR experiment (not observable) could produce a significant increase in the helical population.
It is important to consider that the presence of aromatic residues can significantly affect the far-UV CD spectrum of your peptide, as well as the chemical shifts of the CaH protons in the surrounding residues. Therefore, the comparison with the experimental data is less reliable.