<s>
PSI-blast	O
based	O
secondary	O
structure	O
PREDiction	O
(	O
PSIPRED	B-Algorithm
)	O
is	O
a	O
method	O
used	O
to	O
investigate	O
protein	O
structure	O
.	O
</s>
<s>
It	O
uses	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
machine	O
learning	O
methods	O
in	O
its	O
algorithm	O
.	O
</s>
<s>
It	O
is	O
a	O
server-side	B-Application
program	O
,	O
featuring	O
a	O
website	O
serving	O
as	O
a	O
front-end	O
interface	O
,	O
which	O
can	O
predict	O
a	O
protein	O
's	O
secondary	O
structure	O
(	O
beta	O
sheets	O
,	O
alpha	O
helixes	O
and	O
coils	O
)	O
from	O
the	O
primary	O
sequence	O
.	O
</s>
<s>
PSIPRED	B-Algorithm
is	O
available	O
as	O
a	O
web	O
service	O
and	O
as	O
software	O
.	O
</s>
<s>
The	O
software	O
is	O
distributed	O
as	O
source	O
code	O
,	O
licensed	O
technically	O
as	O
proprietary	B-Application
software	I-Application
.	O
</s>
<s>
It	O
allows	O
modifying	O
,	O
but	O
enforces	O
freeware	B-License
provisions	O
by	O
forbidding	O
for-profit	O
distribution	O
of	O
the	O
software	O
and	O
its	O
results	O
.	O
</s>
<s>
Secondary	O
structure	O
prediction	O
involves	O
a	O
set	O
of	O
methods	O
in	O
bioinformatics	O
that	O
aim	O
to	O
predict	O
the	O
local	B-Algorithm
secondary	I-Algorithm
structures	I-Algorithm
of	O
proteins	O
and	O
RNA	O
sequences	O
based	O
only	O
on	O
knowledge	O
of	O
their	O
primary	O
structure	O
–	O
amino	O
acid	O
or	O
nucleotide	O
sequence	O
,	O
respectively	O
.	O
</s>
<s>
This	O
matrix	O
is	O
processed	O
by	O
an	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
,	O
which	O
was	O
constructed	O
and	O
trained	O
to	O
predict	O
the	O
secondary	O
structure	O
of	O
the	O
input	O
sequence	O
;	O
in	O
short	O
,	O
it	O
is	O
a	O
machine	O
learning	O
method	O
.	O
</s>
<s>
PSIPRED	B-Algorithm
works	O
to	O
normalize	O
the	O
sequence	O
profile	O
generated	O
by	O
PSIBLAST	O
.	O
</s>
<s>
For	O
each	O
amino	O
acid	O
in	O
the	O
sequence	O
,	O
the	O
neural	B-Architecture
network	I-Architecture
is	O
fed	O
with	O
a	O
window	O
of	O
15	O
acids	O
.	O
</s>
<s>
Added	O
information	O
is	O
attached	O
,	O
indicating	O
if	O
the	O
window	O
spans	O
the	O
N	O
or	O
C	B-Language
terminus	O
of	O
the	O
chain	O
.	O
</s>
<s>
A	O
second	O
neural	B-Architecture
network	I-Architecture
is	O
used	O
to	O
filter	O
the	O
predicted	O
structure	O
of	O
the	O
first	O
network	O
.	O
</s>
<s>
Using	O
the	O
secondary	O
structure	O
with	O
the	O
highest	O
score	O
,	O
PSIPRED	B-Algorithm
generates	O
the	O
protein	O
prediction	O
.	O
</s>
