<s>
In	O
bioinformatics	O
,	O
Stemloc	B-Application
is	O
an	O
open	O
source	O
software	O
for	O
multiple	B-Algorithm
RNA	I-Algorithm
sequence	I-Algorithm
alignment	I-Algorithm
and	O
RNA	O
structure	O
prediction	O
based	O
on	O
probabilistic	O
models	O
of	O
RNA	O
structure	O
known	O
as	O
Pair	B-General_Concept
stochastic	I-General_Concept
context-free	I-General_Concept
grammars	I-General_Concept
(	O
also	O
probabilistic	B-General_Concept
context-free	I-General_Concept
grammars	I-General_Concept
)	O
.	O
</s>
<s>
Stemloc	B-Application
attempts	O
to	O
simultaneously	O
predict	O
and	O
align	O
the	O
structure	O
of	O
RNA	O
sequences	O
with	O
an	O
improved	O
time	O
and	O
space	O
cost	O
compared	O
to	O
previous	O
methods	O
with	O
the	O
same	O
motive	O
.	O
</s>
<s>
Stemloc	B-Application
was	O
written	O
in	O
2004	O
by	O
Ian	O
Holmes	O
.	O
</s>
<s>
Stemloc	B-Application
can	O
be	O
downloaded	O
as	O
part	O
of	O
the	O
.	O
</s>
<s>
A	O
previously	O
developed	O
algorithm	O
by	O
David	O
Sankoff	O
in	O
1985	O
uses	O
dynamic	B-Algorithm
programming	I-Algorithm
to	O
simultaneously	O
align	O
and	O
predict	O
multiple	O
RNA	O
structures	O
.	O
</s>
<s>
This	O
is	O
observantly	O
expensive	O
,	O
and	O
thus	O
is	O
the	O
motivation	O
to	O
create	O
better	O
RNA	O
analysis	O
tools	O
like	O
Stemloc	B-Application
.	O
</s>
<s>
The	O
initial	O
goal	O
of	O
Stemloc	B-Application
was	O
to	O
reduce	O
the	O
time	O
and	O
space	O
cost	O
of	O
simultaneous	O
alignment	O
and	O
structure	O
prediction	O
of	O
two	O
RNA	O
sequences	O
by	O
using	O
a	O
stochastic	B-General_Concept
context-free	I-General_Concept
grammar	I-General_Concept
(	O
SCFG	O
)	O
scoring	O
scheme	O
and	O
by	O
implementing	O
constrained	O
versions	O
of	O
the	O
Sankoff	O
Algorithm	O
.	O
</s>
<s>
Stemloc	B-Application
uses	O
alignment	O
envelopes	O
and	O
fold	O
envelopes	O
to	O
simultaneously	O
constrain	O
both	O
the	O
alignment	O
and	O
the	O
secondary	O
structures	O
of	O
the	O
sequences	O
being	O
compared	O
.	O
</s>
<s>
Both	O
global	O
and	O
local	B-Algorithm
alignment	I-Algorithm
is	O
supported	O
.	O
</s>
<s>
Input	O
in	O
Stemloc	B-Application
can	O
either	O
be	O
in	O
FASTA	O
or	O
Stockholm	O
format	O
(	O
see	O
above	O
for	O
descriptions	O
of	O
each	O
)	O
.	O
</s>
<s>
Sample	O
input	O
shown	O
below:The	O
"	O
--local	O
"	O
command	O
analyzes	O
the	O
file	O
in	O
local	B-Algorithm
alignment	I-Algorithm
mode	O
.	O
</s>
<s>
Stemloc	B-Application
relies	O
heavily	O
on	O
stochastic	B-General_Concept
context-free	I-General_Concept
grammars	I-General_Concept
,	O
which	O
can	O
be	O
seen	O
as	O
a	O
scoring	O
scheme	O
for	O
the	O
algorithm	O
.	O
</s>
<s>
To	O
better	O
this	O
,	O
Stemloc	B-Application
allows	O
the	O
user	O
to	O
constrain	O
the	O
total	O
number	O
of	O
folds	O
and	O
alignments	O
to	O
be	O
considered	O
.	O
</s>
<s>
The	O
main	O
idea	O
of	O
Stemloc	B-Application
is	O
being	O
able	O
to	O
set	O
a	O
threshold	O
for	O
the	O
number	O
of	O
folds	O
and	O
alignments	O
that	O
are	O
sampled	O
to	O
create	O
the	O
envelopes	O
.	O
</s>
<s>
Another	O
feature	O
of	O
Stemloc	B-Application
is	O
its	O
ability	O
to	O
parameterize	O
probabilistic	O
models	O
like	O
stochastic	B-General_Concept
context-free	I-General_Concept
grammars	I-General_Concept
from	O
data	O
.	O
</s>
<s>
Stemloc	B-Application
utilizes	O
the	O
Inside-Outside	B-Application
algorithm	I-Application
and	O
stochastic	B-General_Concept
context-free	I-General_Concept
grammars	I-General_Concept
to	O
maximize	O
the	O
likelihood	O
of	O
a	O
training	O
set	O
.	O
</s>
<s>
This	O
is	O
useful	O
because	O
the	O
default	O
parameters	O
for	O
Stemloc	B-Application
were	O
trained	O
on	O
a	O
selection	O
of	O
pairwise	O
alignments	O
of	O
between	O
30%	O
and	O
40%	O
sequence	B-Algorithm
identity	I-Algorithm
from	O
Rfam	O
(	O
database	O
)	O
version	O
5.0	O
.	O
</s>
<s>
Stemloc	B-Application
has	O
since	O
been	O
used	O
in	O
a	O
variety	O
of	O
research	O
publications	O
in	O
RNA	O
structure	O
analysis	O
.	O
</s>
<s>
Most	O
notably	O
in	O
the	O
study	O
of	O
optimal	O
multiple	O
sequence	B-Algorithm
alignment	I-Algorithm
.	O
</s>
