Divide and Conquer Strategies for Protein Structure Prediction - Mathematical Approaches to Polymer Sequence Analysis and Related Problems - page 28

are available (MAXHOM [ 44 ], CLUSTALW [ 47 ], T-Cofee [ 29 ]andMUSCLE[ 10 ]
are currently the most widely used) and none of them is globally accepted as a
standard.
Few years ago, the procedure described above for building MSA was almost
standard and time-consuming; thus, during the construction and tuning of new
prediction methods most of the researchers used the homology-derived secondary
structure of proteins (HSSP) precompiled multiple sequence alignments generated
with the MAXHOM software. Currently, a faster and more accurate method for
the construction of reliable sequence profiles is the adoption of the position spe-
cific iterative (PSI) feature in BLAST [ 2 ]. In PSI-BLAST a sequence profile and
a position-specific scoring matrix (PSSM) are automatically constructed from a
pseudo-multiple alignment of the highest scoring hits in an initial BLAST search.
The PSSM is generated by calculating position-specific scores for each position
in the alignment. Highly conserved positions receive high scores and weakly con-
served positions receive scores near zero. The profile is used to perform a further
BLAST search and the current profile is refined according to the outcomes of the
new search. This iterative procedure is performed until the retrieved sequences re-
main constant or a fixed number of iterations are achieved. In [ 21 ], the prediction
accuracy of secondary structure was improved by using directly the PSI-BLAST
PSSM to feed a neural network system.

2.4 Secondary Structure Prediction

In biochemistry and structural biology, the protein secondary structure refers to
the three-dimensional shape of consecutive residue segments. The most common
secondary structure elements are alpha-helices and beta-sheets. The formation of
secondary structure elements is mostly guided by local inter-residue interactions
mediated by hydrogen bonds. For example, an alpha-helix is formed when hydro-
gen bonds occur regularly between positions i and i C 4 in a protein segment. When
hydrogen bonds occur between positions i and i C 3,thena3 10 helix is formed. A
beta-sheet is formed when two strands are joined by hydrogen bonds involving alter-
nating residues on each participating strand. In the 1950s, Pauling correctly guessed
the formation of helices and strands [ 31 , 32 ], before any protein structure had been
determined experimentally.
There are several methods for defining protein secondary structure elements. The
dictionary of protein secondary structure (DSSP) method [ 23 ] is actually considered
the de facto standard for secondary structure definition. The DSSP defines eight
types of secondary structure elements, based on hydrogen-bonding patterns as those
initially proposed by Pauling (Fig. 2.2 ):

G D 3-turn helix (3 10 helix). Min length three residues.

H D 4-turn helix (alpha helix). Min length four residues.

I D 5-turn helix (pi helix). Min length five residues.

T D hydrogen bonded turn (3, 4 or 5 turn).