Skip to main content

Table 1 Programs used in MESSA for prediction of local sequence features and their interpretation

From: MESSA: MEta-Server for protein Sequence Analysis



Programs used


Secondary structure

Assist three-dimensional structure and domain boundary prediction.

PSIPRED (v2.0) [49]

SSPRO (v4.0) [50]

DISEMBL (v1.5) [51], coils

PSIPRED and SSPRO predict 3-states secondary structures (H: α-helix, E: β-strand, C: coils); DISEMBL predict coils (lower-case letters highlighted in pink)

Disordered and flexible region

Assist three-dimensional structure prediction.

DISEMBL (v1.5) [51], hot loops

Loops that are likely to have high B factors in the X-ray crystallography (lower-case letters highlighted in pink)


DISEMBL (v1.5) [51], missing

DISPRO (v1.0) [52]

DISOPRED (v2.0) [53]

IsUnstruct (v2.02) [54]

Residues without a defined structure (represented by star marks and highlighted in red)

Transmembrane helix and Signal Peptide

Predict subcellular localization and transmembrane, reveal topology of transmembrane proteins and provide hints to the protein function.

TMHMM (v2.0) [55]

TOPPREDa (v2.0) [56]

HMMTOPa (v2.0) [57]

MEMSAT (v3.0) [58]

H: transmembrane helix (colored in blue); h: not confidently predicted transmembrane helix; o: periplasmic loop, i: cytoplasmic loop. x: loop region (not specified as periplasmic or cytoplasmic).



Phobius [60]

H: transmembrane helix (colored in blue); S: signal peptide (colored in green); h: unconfident transmembrane helix; o: periplasmic loop, i: cytoplasmic loop.


SignalP (v3.0) [61] (HMM mode)

SignalP (v3.0) [61] (NN mode)

S: signal peptide (highlighted in green) o: periplasmic region; x: do not have signal peptide

Low-complexity region

Reveal false positive hits of homology search caused by matching of low-complexity region.

SEG [62]

The part with low diversity in amino acid composition (highlighted in pink), likely to be disordered or fold as α helices, such as coiled coil

Coiled coil

Assist three-dimensional structure prediction.

COILS [63]

x: coiled coils, highlighted in yellow

Conservation index

Reveal essential residues for the folding and function of a protein.

BLAST (hits filtered by > 40% coverage and < 90% identity are included in the profile), AL2CO (calculate conservation indices based on profile) [64]

Sequence highlighted by the conservation (highlighted from white, through yellow to dark red as conservation increases)

  1. aTOPPRED and HMMTOP are mainly designed to predict the topology of a given membrane protein rather than distinguish transmembrane proteins from cytoplasmic ones. Thus they may recognize the hydrophobic buried helices in cytoplasmic proteins as transmembrane helices, leading to a high false positive rate.