In this work, recently solved crystal structures of membrane proteins are e
xamined with respect to the performance of the Web server SPLIT in predicti
ng sequence location, conformation and orientation of membrane associated p
olypeptide segments. The SPLIT predictor is based on the preference functio
ns method. Preference functions serve to transform the input choice of amin
o acid attributes into sequence dependent conformational preferences. Trans
membrane helical segments are accurately predicted with a good selection of
preference functions extracted from the compiled database of non-homologou
s integral membrane proteins. Unlike other algorithms with similar high acc
uracy, the SPLIT predictor requires no homology information. With preferenc
e functions extracted from soluble proteins, the sequence location of short
er non-transmembrane helices can be also found in membrane proteins. In par
ticular, Richardson's preference functions are even better than hydrophobic
moments in finding interface helices at the water/lipid phase boundary. Th
e Internet access for the SPLIT system is at the address: http://pref.etfos
.hr/split.