Database Reference
In-Depth Information
Table 1. Identification of WT-TTR residues present in the most prevalent correlations among all com-
puted clusters. The five data sets are identified as Run 1 to 5, corresponding to the five MD unfolding
simulations. Each gray row defines the set of residues under consideration. Properties P1, P2 and P3
correspond to the average 1D distance (number of residues), the average 3D distance (Å) and the aver-
age hydrophobicity (kcal/mol), respectively, for the cluster considered
Run 1
Run 2
Run 3
Run 4
Run 5
Phe33, Lys70
(P1) Average 1D distance
41.88
28.06
42.12
35.50
20.25
(P2) Average 3D distance
12.70
10.80
24.85
7.42
10.95
(P3) Average hydrophobicity
-0.76
-0.06
-1.75
-1.76
-1.11
Ala36, Asp39, Glu42
(P1) Average 1D distance
14.09
2.50
12.12
4.88
17.19
(P2) Average 3D distance
17.00
16.17
17.12
29.81
14.92
(P3) Average hydrophobicity
-5.36
-2.40
-0.57
-7.60
-3.03
Leu12, Val14, Val71, Ile107, Ala109, Leu111
(P1) Average 1D distance
42.06
7.62
38.50
13.25
49.00
11.24
45.69
(P2) Average 3D distance
10.73
4.43
4.83
7.49
11.19
5.63
9.41
(P3) Average hydrophobicity
2.86
1.00
4.26
4.39
0.16
0.16
2.88
of residues co-occurring in the same clusters
across multiple data sets, one could speculate on
the cooperative role of the correlated residues in
the unfolding process of the protein WT-TTR.
In order to find groups of residues conserved in
the same clusters across multiple data sets, an
itemset mining algorithm was applied (Agrawal
& Srikant, 1994). This type of algorithms allows
the discovery of elements that co-occur a number
of times equal or greater than a threshold value
- minimum support. In the context of this work,
minimum support corresponds to the minimum
number of data sets for which a group of residues
is expected to co-occur.
After comparing the clusters of the five data
sets, we discovered two groups of residues with
prevalent correlations in all data sets: (i) Phe33
and Lys70, and (ii) Ala36, Asp39 and Glu42. Ad-
ditionally, we found ten pairs of residues clustered
together in four of the five data sets, with most of
the pairs involving at least one of the following
hydrophobic residues: Leu12 and Val14 (β-strand
A), Val71 (β-strand E), and Ile107, Ala109 and
Leu111 (β-strand G). Table 1 presents the proper-
ties P1, P2 and P3 of the clusters containing the
groups of residues identified above, for each data
set (Run 1 to Run 5).
For the clusters containing residues Phe33
and Lys70, the data show that the average linear
distance between residues (P1) in these clusters
is high which indicates that stronger correlations
appear first between residues far apart in the
protein linear sequence; however the residues
seem to appear in close spatial proximity (low
values for property P2). Moreover, the residues
composing these clusters are mainly hydrophobic,
with most of them positioned in β-strands B, C
and E. As for the properties characterizing the
clusters containing residues Ala36, Asp39 and
Glu42, for Runs 1, 3 and 5 higher correlation
values are observed between residues far apart in
the protein linear sequence, whereas for Runs 2
and 4 precisely the opposite happens. The values
of average spatial distance (P2) are more or less
of the same order of magnitude in all data sets,
except for Run 4, where residues are close in the
Search WWH ::
Custom Search