Experiments and Results - Protein Homology Detection Through Alignment of Markov Random Fields - page 39

reference alignment that are aligned by an algorithm. Alignment precision is

de ned as the fraction of aligned positions that are correct according to a reference

alignment. Since reference alignments are used to judge if one residue is correctly
aligned or align-able, their quality is critical. To reduce bias, we use three very
different structure alignment tools TM-align [ 9 ], Matt [ 10 ], and DeepAlign [ 11 ]to
generate reference alignments.

4.3 Reference-Dependent Alignment Recall

As shown in Tables 4.1 and 4.2 , our method MRFalign exceeds all the others
regardless of the reference alignments on both dataset Set3.6K and Set2.6K. When
only exact matches are considered as correct, MRFalign outperforms HHalign by
*

24 % on Set3.6K and
Set2.6K, respectively. If 4-position off exact match is allowed in calculating
alignment recall, MRFalign outperforms HHalign by

10 % on both datasets, and HHMER by * 23 and *

* 11 % on both datasets, and

HHMER by 33 %, respectively.

On the very large set Set60K, as shown in Tables 4.3 , 4.4 and 4.5 , MRFalign
outperforms the other two at each SCOP classi

* 25 and *

cation level regardless of the
employed reference alignments. MRFalign is only slightly better than HHalign at
the family level, which is not surprising since it is easy to align two closely-related
proteins. At the superfamily level, MRFalign outperforms HHalign and HMMER

Table 4.1 Reference-dependent alignment recall on Set3.6K
TMalign

Matt DeepAlign

4-offset
(%)
HMMER 22.9 26.5 24.1 27.4 25.5 28.1
HHalign 36.3 39.1 37.0 42.1 38.4 42.8
MRFalign 47.4 51.0 47.5 52.6 49.2 53.5
Three structure alignment tools (TMalign, Matt and DeepAlign) are used to generate reference
alignments. “ 4-offset ” means that 4-position off the exact match is allowed. The bold indicates the
best results

Exact
match (%)

4-offset
(%)

Exact
match (%)

4-offset
(%)

Exact
match (%)

Table 4.2 Reference-dependent alignment recall on Set2.6K
TMalign Matt DeepAlign

Exact
match (%)

4-offset
(%)

Exact
match (%)

4-offset
(%)

Exact
match (%)

4-offset
(%)

HMMER 36.5 42.6 38.6 44.0 40.4 45.0

HHalign 62.5 66.1 63.2 66.2 64.0 66.7

MRFalign 72.8 76.2 73.5 76.7 74.2 77.8