►
Description
Open project meeting for Open Source Antibiotics Series 1, the Mur Ligases.
Full Project: https://github.com/opensourceantibiotics/murligase
Relevant GitHub Issue: https://github.com/opensourceantibiotics/murligase/issues/47
On the call: Professor Matthew Todd, Dr Dana Klug, Yuhang Wang (UCL), Professor Chris Dowson, Laura Diaz Saez, Rebecca Steventon (University of Warwick), Lizbe Koekemoer (Diamond/Oxford), Dr Lori Ferrins, Dr Joe Eyermann (NEU), Peter Horanyi (UCB), Dr Jan Jensen (University of Copenhagen).
A
Okay,
welcome
to
the
open
source,
antibiotics,
my
ligase
meeting-
this
is
tuesday
6th
of
july
at
2pm,
thanks
for
coming
along
and
the
agenda
and
actions
were
posted
last
time.
Sorry
just
posted
last
night
and
I
will
just
bring
those
up
so
that
we're
all
looking
the
same
thing
in
case
people
don't
have
in
front
of
them,
and
that
is
going
to
be
hopefully
that
page.
A
So
there
are
a
bunch
of
things
here
which
you
know
are
sort
of
sitting
here
as
as
to
do
items,
but
I
guess
there
were
a
few
things
that
I
just
wanted
to
try
and
cover
first
so
just
on
the
shipment
of
the
atomized
compounds,
which
we're
all
very
excited
to
test.
So
we
do.
We
do
have
this,
this
approval
leadership
to
diamond,
but
we
were
holding
out
for
agreement
to
ship
directly
to
the
warwick
first
eula.
A
I
think
to
so
that
you
could
have
a
compound
and
then
maybe
take
them
to
diamond,
and
we
are,
I
think,
very
close
to
getting
that.
The
document
is
back
with
ucl's
legal
team
and
I'm
hoping
that
they
can
just
agree
to
warwick's
queries
and
questions,
and
then
we
can
get
going
in
the
meantime
dana's
taking
a
look
at
some
of
the
compounds,
a
sample
of
the
compounds
and
they
look
fine.
A
So
we
don't
have
any
issues
to
do
with
things
going
off
because
they've
been
kept
properly.
So
I
think
we're
still
fine
and
I
was
going
to
give
the
ucl
team
just
a
day
or
so
and
then
and
then,
if
we
can't,
if
there's
any
issues
with
the
remaining
paperwork,
I
was
gonna.
A
I
was
gonna,
maybe
try
and
think
of
a
way
around
this,
like
maybe
shipping,
the
diamond
first,
even
though
that's
that's
like
the
plan
b,
but
certainly
in
the
short
term,
we'll
I
think
we'll
resolve
this
one
way
or
the
other.
That's
where
we
currently
stand.
Nothing
really
for
anyone
to
do
there.
It's
just
that.
A
You
know
I'm
keen
to
sort
of
get
this
going
in
part,
because
we've
had
just
a
new
input
in
the
last
few
days
from
yen
who's
with
us,
who's
been
working
on
docking,
some
of
the
some
commercial
compounds,
eventually
the
available
compounds
into
mercy.
So
I
was
hoping
we
could
maybe
start
with
with
this,
because
it's
it's
kind
of
relevant
in
the
sense
we're
again
talking
about
you
know,
aim
id
commercially
available
compounds
which
which
appear
to
be
scoring
well.
A
So
you
posted
everything
here
again
I
mean:
did
you
want
to
talk
about
a
little
bit
and
just
bring
everyone
up
to
speed
in
case
they
haven't?
Seen
what's
been
going
on
here?
Yes,.
B
I
just
wanted
to
sort
of
it's
basically
more
where
we're
coming
from
so
so
so
casper
steinmann
and
I
wrote
a
paper
a
couple
of
months
ago,
where
we
use
kinetic
algorithms,
basically
to
create
new
molecules
that
have
good
docking
scores,
and
so,
as
part
of
that,
we
also
compare
the
docking
scores
to
commercial
databases,
just
to
make
the
point
that
we
can
actually
get
significantly
better
docket
scores,
and
so
that's
typically.
B
The
first
thing
we
do
is
we
sort
of
know
what
docking
scores
to
beat,
and
so
we've
done
that,
and
I
just
thought
I
would
put
those
those
up.
So
the
genetic
algorithm
is
running
right
now
and
we'll
get
other
molecules
that
have
significantly
better
docking
scores
but,
of
course,
won't
be
commercially
available.
Unless
we,
unless
it
sort
of
gets
lucky,
you
can
see
a
question
here.
C
Oh
yeah,
thanks
hans,
it's
christos
yan
from
warwick
university.
It's
great
that
you're
engaging
with
this
brilliant,
I'm
just
wondering
as
part
of
your
algorithms,
whether
you
can
put
into
this
it's
something
like
the
entryway
program.
That's
used
to
understand
getting
drugs
into
the
gram
negatives
as
well.
I
I'd
see
a
lot
a
lot
of
hedge
cycles,
I'm
just
wondering
you
know
the
relative
hydrophobicity
of
these.
B
Yeah,
so
I
mean
one
thing
we
did
to
check
was:
was
just
to
log
p
values.
So
if
you
go
into
the
notebook,
you
can
sort
of
put
your
mouse
over
each
molecule
and
you
get
the
log
p
value
and
they
all
seem
reasonable
in
in
in
principle
anything
that
we
can
predict
relatively
quickly.
B
A
A
To
a
a
set
of
sort
of
heuristics
about
what
makes
molecules
accumulate
in
gram-negative
bacteria,
so
features
like
primary
amines
and
things
like
that
and
whether
you
can
apply
and
and
globularity
right,
so
things
that
work
best.
If
there's
a
way
of
applying
a
set
of
rules
on
top
yep.
B
So
we
so
we
can
do
that
if
we
have
access
to
the
the
software
package.
Basically
so
so,
if
someone
can
put
details
up
on
the
on
the
the
issue
there
on
the
channel,
but
I.
D
B
B
Yeah,
it's
just
sort
of
the
you
know
in
the
course
of
a
genetic
algorithm
run
we
screen
about
400
000,
so
it
has
to
be
fairly
quick.
So
so,
basically,
what
I
I
just
need.
Is
it
a
commercial
piece
of
software?
Is
it
something
we
can
download.
A
E
B
B
So
there
are
two
options:
one
is
to
do
a
post
screen
or
a
prescreen.
Actually,
if,
if
it
turns
out
that
the
I
mean,
if
you're
doing
it
for
a
database
right,
we
can,
we
can
well
yeah.
So
so
let
me
back
up
if,
if
it's,
if
it's
something
where
you
have
to
sort
of,
if
where
you
can
run
natively
where
you
have
to
call
and
rely
on
network
traffic,
then
we
can't
it's
for
practical
reasons,
it's
hard
to
do
on
the
fly.
C
So
I
wonder
whether
it's
worth
a
discussion
with
paul
paul
hergenrother's
group
in
illinois
who
who
wrote
that
entryway
program
and
and
maybe
he
might
be
prepared
to
open
up
the
code
and
maybe
think
about
being
able
to
integrate
it
as
a
single
package
rather
than
yeah
yeah.
C
And
I
I
I
he
seems,
I
think
he
might
be
that
person
who
might
be
keen
on
that
yeah,
yeah
and
and
yeah
that'd
be
a
good
start
point
yep
yeah.
B
I
can
definitely
look
into
it.
I
mean
sort
of
the
yeah.
If
we
can
get
the
number
quickly,
then
it's
not
a
problem
to
include
in
the
genetic
algorithm.
So
we
already
have
things
in
genetic
algorithm,
that
screens
for
synthetic
accessibility
and
things
like
that,
and
we
also
almost
put
in
the
log
p,
but
it
just
turned
out
that
we
get
molecules
with
reasonable
log
p
values
anyway,
but
we
could
also
throw
out
high
luck.
P,
values,
if
you
want.
A
B
So
so
it's
so
I
did
a
little
check
now
on
the
actual
availability.
So
so,
basically
we
weren't
very
thoughtful
about
picking
the
we
picked
the
enemy
because
we
knew
those
were
purchasable
and
the
the
list
of
zinc
compounds
we
just
had
lying
around
from
from
other
stuff,
so
these
are
not
necessarily
in
stock.
I
just
before
the
meeting
I
just
checked
the
top
ten.
B
With
this.
The
mole
part
program
that
was
suggested
and
only
number
two
and
four
are
are
actually
in
stock
from
enemy.
I
could
see
so
I
and
I
can.
I
can
check
the
other
ones
too,
so
it
it's.
I
mean
if
these
are
really
so,
I
guess
they're
with
the
libraries.
B
There
are
several
options
I
mean
there
are
it's
possible
to
just
use,
in-stock
molecules
from
zinc
and
I'm
actually
right
now,
I'm
trying
to
download
that
set
of
compounds
and
also
there
are,
I
think,
the
reason
that
the
sink
molecules
score
better
than
the
enemy
is.
We
used
a
small
enemy
instead
of
only
10
000
and
a
large
sink
set
of
250
000,
but
there
is
an
enemy
in
set
of
250
000.
So
we
could
also
try
to
dock
that
instead.
B
So
I
think
in
principle
the
as
far
as
I
know,
the
the
enemy
status
should
be
all
purchasable,
maybe
not
in
stock,
but
maybe
you
know
sometimes
to
make
them
on
demand.
A
B
Yeah,
that's
that's.
The
other
thing
we
can
do
is
so
we
have.
We
can
interface
this
with
the
retrosynthesis
program
and
basically
give
you
for
each
molecule
the
the
minimum
number
of
synthetic
steps
for
each.
So
that's
that's
something
I
can
add
fairly
easily
to
the
list
and
then
we
can
rank
them
based.
A
B
If
you're
doing
a
lot
of
them
right,
then
yeah,
but
but
then
I
think
it
might
be
worth
waiting
for
the
genetic
algorithm
once
because
they
will.
They
will
definitely
have
a
higher
docking
score.
I
mean,
if
you
have
to
synthesize
anyway,
they
may
they
will
definitely
have
a
higher
docking
score
and
maybe
will
be
equally
easy
to
make
okay.
So
I
think
those
will
come
next
week,
sometime
they're
running
right
now,.
B
Also,
I
just
from
sort
of
from
from
my
point
of
view.
I
mean
it
would.
Of
course
we
want
to.
We
want
to
help
in
the
effort,
but
but
you
know
we're
basically
where
we're
coming
from.
Is
we
if
we,
if
we
could
get
some
experimental
validation
on
the
genetic
algorithm
predictions,
and
that
would
that
that
would
be
wonderful
for
us,
yeah,
absolutely
yeah,
completely
understood
and
that's
what
we
want
to
yeah
and
that
could
also
be
a
paper.
B
A
Yeah
no
awesome
so
I
mean
just
on
the
I
guess
there
are
two
follow-up
questions.
One
is
that
I
assume
that
you
know
how
these
molecules
are
being
predicted
to
bind
so
the
pose
in
the
enzyme.
Is
that
right?
Okay,
and
so
I'm
assuming
that
if
we
could
cluster
them,
we
might
learn
something
about
the
value
of
certain
residues
as
binding
partners
for
small
molecules.
B
Yes,
so
I
I
so
the
the
the
person
was
actually
doing
the
docking.
That's
my
collaborator
casper
he's
on
vacation
this
week,
but
I
mean,
if
I
don't
know
if
he
saved
them
some,
because
we
did
400
000
per
run
right
so,
but
if,
if
not,
then
we
could
certainly
do
the
let's
say
the
top
thousand
ones
again
to
just
redock
them
and
then
this
time
save
the
save
the
pros.
A
Okay,
great,
I
mean,
I
guess,
I'm
kind
of
interested
in
the
3d
shape
of
what
they
make
as
the
ensemble.
You
know,
and
then
the
second
thing
was
so.
A
lot
of
this
project
is
based
on
the
idea
of
one
molecule
hitting
more
than
one
enzyme,
a
so-called
multi-targeting
approach.
So
I
was
wondering
if
it's
possible,
with
the
results
of
the
genetic
algorithm
when
it's
finished
to
run.
You
know
the
sort
of
top
hundred
against
one
of
the
other
muraling
enzymes
chris,
I'm
assuming.
B
Yeah
yeah,
we
could
certainly
do
that
as
long
as
we
have
the
x-rays
or
any
structure,
and
we
could
also,
I
mean,
I
think,
that
the
the
re-docking
of
the
top
100
or
whatever
is
easiest,
but
but
if
that
turns
out
to
be
sort
of
a
promising
direction,
it's
also
possible,
in
principle
at
least,
to
to
dock
both
doing
the
genetic
algorithm
and
sort
of
steer
the
the
search
towards,
let's
say,
binding
one,
but
not
the
other
or
binding,
both
or
whatever
you're
most.
C
Right
right,
right,
right
up
front
across
a
pair
of
ligases,
it
depends
really
on
how
much
computing
power,
but
you
know
in
an
ideal
world,
you'd
be
looking
for
a
pair
of
ligases
and
then
picking
off
several
species
simultaneously.
That
could
form
a
you
know,
a
clinical
indication
that
would
be
your
target
product
profile.
C
You've
got
okay,
you
can
go
for
your
retract
infections
or
respiratory
infections
and
you're
immediately.
You
know
focusing
your
screen
to
give
you
the
best
hit
against
the
two
or
three
pathogens
that
are
the
most
important
and
and
also
multiply
targeting
so
small
ask,
but
that
would
be
lovely
yeah.
B
B
B
Very
target
specific,
actually,
you
know
some
some
targets,
the
scoring
function,
at
least
in
glide,
which
is
what
we're
using
seems
to
be
picking
up
the
the
physics
and
then
some
targets
it
just
doesn't.
We
we
tried
something
similar
with
the
the
covet
main
protease,
where
they
then
generated
a
lot
of
data,
and
there
is
absolutely
no
correlation
between
the
docking
score
and
the
ic50s
on
that
system.
So
all
right,
that's
brilliant!.
F
Yeah
I
was
going
to
ask
about
the
I
see
a
number
of
compounds
that
are
very
similar.
Looking
to
me
with.
E
F
B
I
can
certainly
I
mean,
what's
very
easy
to
do-
is
just
calculate
a
similarity
score
sort
of
between
them
so
so
and
then
sort
of
set
a
cut
off.
I
think
the
usual
sort
of
cutoff
and
drug
discoveries-
I
don't
know
either
0.6
or
0.8.
You
know
where
one
is
the
perfect
match,
so
we
could
sort
of
indicate
if
there
are
other
higher
scoring
ones.
So
we.
F
Yeah
yeah,
that
makes
sense,
I
mean
I'm
just
saying
that,
like
there
are
many
of
them,
numerous
ones
here,
like
the
top
left,
two
are
very
similar.
The
warhead
is
similar.
The
sort
of
the
the
linkages
are
different,
but
I
just
see
like
a
number
of
them
yeah,
the
ones
your
cursor
is
on.
Those
two
are
also
yes.
I
just
think
that
you
know.
B
Yeah
the
question
is,
we
can
certainly
do
that.
I
should
just
I
should
say:
I'm
not
really
a
med
chemist.
So
the
question
is
at
what
point
will
I
make?
What
I'm
worried
about
is
sort
of
making
stupid
decisions
and
you're
never
seeing
the
molecules.
So
at
what
point
sort
of
do,
the
experts
come
in
and
just
sort
of
you
know
throw
them
out.
A
And
at
what
point
do
I
do
it?
No,
I
think
peter
I
mean
peter's
worried
about
you
know
doing
crystallography.
Let's
say
on
some
of
these
and
instead
of
doing
15,
you
want
to
pick
five.
Do
you
want
just
because
the
experiments
are
quite
labor
intensive,
so
he's
thinking?
Well,
let's
try
five
diverse
compounds,
but
then
it's
probably
best
that,
like
an
expert
picks.
B
Them
I
mean.
F
A
A
B
I
can
I
can
look
into
how
to
make
that
and
try
to
make
that
it
sort
of
depends.
I
mean
if,
if
I
guess
you
know,
if,
if
you
have
to
sift
through
a
thousand,
then
it's
probably
best
that
I
code
something
up.
If
you
have
to
sift
through
100
to
pick
five,
I
would
probably
just
eyeball
it.
I
mean
it's
just
safe
to
do
that.
Yeah.
A
Okay,
fantastic,
okay,
we'll
we'll
wait
the
outcome
of
the
of
the
final
search,
that's
going
on
there
and
then
we'll
make
some
decisions
about
whether
to
make
or
buy
and
what
we're
going
to
do.
But
this
is
an
amazing
start
and
very
exciting
because
of
course
it's
a
fourth
strand
to
what
we
have
here.
It's
a
bunch
of
new
molecules
that
we
could.
We
could
get
access
to
and
start
evaluating.
B
Yep
and
and
at
some
point
when
there's
enough
data
for
for
machine
learning,
then
we
could
also
be
happy
to
make
some
machine
learning
models,
basically
trained
against
ic50s
or
some
experimental
data.
Okay,
that
might
be
more
reliable
than
the
than
the
genetic
algorithm
just.
C
That's
great
and
just
thinking
about
your
tree,
not
just
for
these
compounds,
look
at
their
chemical
relatedness,
but
for
all
of
the
different
compounds
across
all
of
the
different
start
points
we
have.
I
I'm
just
you
know
thinking
you
know
I
I
feel
a
bit
paddington
bare.
My
brain
is
too
small
to
hold
all
these
things
in
you
know
we
need.
You
know
a
a
graphic
that
you
we
could.
We
could
assemble
and
display
all
of
these,
and
particularly
when
we
start
getting
synthetic
derivatives
from
them.
So
we
can.
C
B
It's
it's,
it
sort
of
depends.
If
I,
if
someone
like,
I
said
it's
not
my
field,
but
if
someone
knows
of
a
package
I
can
also
google
around
and
see
what
I
can
find
I
can
talk
to.
Probably
chris
smayne
would
be
a
good
person
to
talk
to
about
this,
and
if
it's
something
open
source
that
we
can
just
you
know
that's
easy
to
download,
then
we
can
certainly
try
so
so
I
just
if,
if
you,
if
you
guys
give
me
inputs
or
suggestions,
then
I
can,
I
can
try
it
out.
B
I
don't
think
I
have
the
background
to
come
up
with
my
own
ideas,
but
I
I
you
know,
I
know
rd
clipped
pretty
well
so
for.
A
Visualization
we
usually
use
data
warrior
right.
If
we've
got
a.
If
we've
got
a
spreadsheet
of
molecules,
we
want
to
see
them,
then
we
use
that
which
generates
something.
That's
not
unlike.
What's
on
the
screen
here
but
yeah,
I
should
also
say:
if
you
go
to
the
notebook,
maybe
I'll,
just.
B
A
B
So
so
it
is
just
have
to
put
it
down
you
can
you
can
see
it
now
right
so
so
it
is
somewhat
interactive,
so
you
can
see
I
can
like
put
data
on
top
right
and
they're.
Also,
these
are
just
the
first
20
right,
but
but
one
can
scroll
through
one
can
also
search
or
sort
by
anything.
B
G
Yeah,
yeah,
okay,
and
just
just
one
small
question:
how
just
about
this
project?
How
many
compounds
are
we
gonna
screen?
Actually
for
you
know
biologically
so
so
we
can
get
enough
data
for
the
machine
learning
generation
to
you
know,
because
the
the
genetic
generic
prediction
is
well
will
be
validated
only
if
we
could
get
enough
data
to
to
to
to
right
measure
the
success
rate
like
what
what
kind
of
predictions
can
be
useful.
What
can
what
kind
of
predictions
are?
B
Yeah
I
mean
for
if,
for
for
machine
learning,
sort
of
the
the,
I
would
say,
the
absolute
minimum
is
maybe
50
and
that
that's
really
dicey
right.
A
hundred
is
a
little
better
but
but
sort
of
as
a
validation
for
the
docking.
I
mean
it.
E
B
G
B
G
Because
we
were
talking
about
like
using
using
selecting
compounds
based
on
their
predicted,
docking
score
and
log
p,
I
think
that's
log
d
right,
so
so
oh
yeah
yeah.
So
so
I
was
wondering
like
do
you?
Do
the
visual
selection
at
the
end?
Or
do
you
just
select
all
by
their
scores?.
B
There
is
some
synthetic
accessibility-
well
actually
not
in
this
one,
since
these
are
purchasable,
but
but
in
the
genetic
algorithm
there
would
be
synthetic
accessibility
for
the
purchasable
ones.
Here
is
just
plain
plain
old,
docking
score,
because
we
just
sort
of
by
visual
inspection.
We
it
looked
like
the
lock
key
was
fine.
I
I
believe
the
rule
of
thumb
is
below
three
and
a
half
or
four.
G
I
saw
you
I
saw
you
published
on
on
the
on
github,
saying
that
you,
you
are
using
the
s
glide
sp,
for
the
crude
virtual
screen
and
then
you
go
for
the
xp
for
more
accurate
predictions.
Yes
and
eventually
selected
okay,
yeah.
G
Do
you
think
we
should
like
try
the
chris
suggestions
like
trying
different
different
algorithms
to
score
the
same
data
sets
like
from
I
mean
from
the
scratch?
Instead
of
just
you
know,
rescore
the
1000,
because
if
we
screen
from
the
scratch,
then
we
can
see
what
what's
actually
overlapping
in
between
and
that
those
might
probably
be.
I
don't
know
if
it's
quite
nice,
but
yeah
I
would
say
I
would
take
that
as
a
bit
more
preferable
source
of
data
yeah.
B
A
lot
in
machine
learning,
also
in
in
general
that
the
main
sort
of
practical
issue
there
is
just
that
I
mean
we
don't
we
have
some.
We
have
experience
with
light.
We
have
some
experience
with
with
something
called
smeena
which
is
like
autograph,
but
we're
not
experts
and-
and
it's
it's
just
hard
to
sort
of
reach,
the
the
level
of
proficiency
required
to
use
the
docking
properly
for
a
bunch
of
different
programs.
B
G
Think,
oh
sorry,
I've
been
thinking
because
ucr
has
like
hpc
resources.
We
can.
We
can
definitely
accelerate
the
docking
process
if
we
can
make
use
of
it
like
the
the
thomas
or
the
yeah
catalina
that
they're
all
available
okay
yeah.
So
I
mean
if
we
can
have
some
software
like
shared
on
this
platform,
so
we
can
just
yeah.
So
we
can
do
a
lot
of
things.
B
So
so
I
mean
all
the
yeah,
basically
all
the
the
smiles
strings
which
you
need,
or
we
can
also
give
structures.
I
mean
these
are
just
I
mean
the
the
both
enemy
and
the
both
data
sets
are
basically
publicly
available
with
the
with
the
genetic
algorithm.
Of
course,
it
would
have
to
be
hooked
up
to
the
genetic
algorithm,
but
maybe
just
the
rescoring
so
running
genetic
algorithm
with
one
algorithm,
docking,
algorithm
and
then
rescoring
it
afterwards
with
other
algorithms
would
also
be
fine.
B
At
and
not
make
stupid
mistakes,
so
I
think
that
would
actually
be
the
rate
limiting
step.
So
I
think
it
would
be
better
if
we
could
get
people
who
are
sort
of
experts
in
the
various
docking
packages
involved
and
they
could
get
our
compound
set,
and
maybe
they
could
give
us
some
compound
sets
that
we
could
compare.
I.
A
Think
that
sounds
very
good.
I
think
we
we
can
take
this
up
on
the
on
the
issue
that
we've
started,
but
that
sounds
very
good
because
you
said
you
know
we
could
just
rescore
things
fairly
easily.
We
can
find
people
willing
to
help
us
out
right.
Peter
did
you
have
a
hand
up
for
historically
or
is
there
another
question.
A
A
Okay,
let
me
just
stick
this
back
up,
which
is
here,
okay,
so
the
I
guess.
The
other
thing
I
was
hoping
to
talk
about
is
so
the
stuff
that
becca
was
doing
with
the
with
both
the
fragment
screen,
where
some
of
the
data
have
been
put
up
on
the
website,
which
is
great
chris,
wanted
the
structures
of
the
molecules
to
be
added.
I'm
not
actually
sure
if
we
have
that,
I'm
assuming
we
have
the
structures
of
those
fragments
somewhere.
Do
we
the
ones
that
were
evaluated.
A
Somewhere
if
they
could
just
be
dumped
somewhere,
if
it's
a
big
file,
then
just
a
sheet,
but
if
it's
a
small
file,
then
structures
or
something
that'd
be
great.
Yes,
okay
and
then,
as
I
understand
it,
there's
been.
There's
been
some
progress
on
the
evaluation
of
dana's
fragments
did
becker
and
danny.
Did
you
want
to
talk
about
these
because
I
think
it's
pretty
hard
for
press.
H
Yeah,
so
we
did
the
initial
screen
with
all
of
them.
Against
my
d
got
a
few
quite
good
hits.
We
got
about
11,
which
reduced
activity
by
about
80
or
more
so
we
double
checked.
There
are
two
inhibitors
by
testing
against
sapling
reaction
and
then
q
inhibitors.
We
tested
against
my
e
from
the
same
bacteria
and
we
appear
to
have
a
couple
which
do
seem
to
be
dual
inhibiting,
which
is
really
good,
no
way.
H
H
Uh-Huh
that
that
would
be
on
dana
to
telling
you
what
they
look
like.
I
worked
off
numbers,
okay,.
E
Oh
so
I
didn't
look
at
all
11,
but
these
are
the
five
six
that
were
tested
against
murray
as
well
as
mer
d,
so
yeah
they.
Actually,
I
don't
know
all
the
fragments
that
we
sent
to
becca
were
fairly
similar,
but
some
of
this
sort
of
extending
off
the
fennel
ring
it
looks
like.
A
H
H
Penetration
of
the
enzyme,
the
conservation
of
the
enzyme
was
2.5
micrograms
per
ml.
H
No,
there
was
definitely
some
which
definitely
were
not
active.
Some
increased
activity
within
the
assay
so.
H
I'd
say
the
I
think
it
was
seven
four
yeah
seven
four
nine
was
mostly
only
targeting
my
d
against
my
e.
There
was
not
a
level
of
inhibition,
we
would
be
expecting.
C
H
Yeah
I
didn't
re-run
the
entire
78
against
my.
A
Very
good
yeah
I'm
liking
that
n-methyl
poverty
wow.
That's
amazing!
That's
on
a
bunch
of
actives,
though
okay
we're
going
to
need
to
digest
those.
I
mean
it's
going
to
be
important
to
compare
the
actives
and
the
inactives
right.
So
if
we,
if
we
find
there
is
a
cluster
there
with
activity
and
and-
and
there
are,
the
same
motif
is
not
found
in
the
inactives,
then
we
are
all
guns
blazing
for
that.
That's
nice.
C
C
B
Yeah,
oh
yeah.
Sorry
I
couldn't
oh
you're
off
on
the
screen.
Just
just
from
experience
I
mean
small
fragments
will
have
low
darkness.
For
us.
This
is
just
yeah.
I
can
say
that
right
now,
but
but
I
mean
there
will
be
sort
of
a
spread
and
in
the
darkness
force.
That
might
be
it's
easier
for
us
to
do.
If
we
get
this,
we
just
need
the
smile
strings
or
the
sdf
file.
C
Is
it
is
it?
Is
it?
Is
it
worth
even
thinking
of
just
getting
docking
data
just
so
we're
having
having
you
know
full
sets
of
data
for
all
the
molecules
that
we're
trying
yeah,
because
we'll
be
doing
the
reverse?
Won't
we
we're
taking
the
best
docking
scores
and
we're
doing
the
biochemical
assays?
I
don't
know,
I'm
happy.
A
H
E
I
Can
you
hear
me
yeah
yeah,
sorry?
Maybe
we
can
give
a
try
for
co-crystallization
with
the
top
compounds,
something
like
that
it
takes
a
while
until
we
get
the
mirror
the
crystals,
that's
like
between
one
and
two
months,
it's
a
very
slow
crystal
system,
so
I
can
throw
more
than
you
know,
maybe
like
at
least
five
compounds.
So
we
can
have
something
for
mere
e
is
faster,
so
that
can
be
another
option
that
yeah.
I
Yeah,
that's
what's
gonna,
be
one
of
my
comments
after
you
know
my
update
for
me
or
d,
I
think
so.
I've
been
trying
to
get
a
new
crystal
system
from
ud
and
it's
not
going
that.
Well,
the
crystals
are
too
small.
They
don't
they're,
not
very
robust
and
in
growing
so
they
grow
a
bit
faster,
but
they
don't
they're
not
consistently
growing.
I
So
I
don't
get
like
a
lot
of
reproducibility
on
the
plate,
which
means
no
feasible
for
x,
cam
or
high
throughput
screening
for
crystals.
So
I
also
was
thinking
that
maybe
we
can
consider
another
organism
for
mere
d
and
I
can
give
it
a
try
with
that
and
see
if
we
can
get
something
faster
and
more
reproducible,
because
waiting
that
long
for
getting
a
crystal
is
gonna,
be
a
bottleneck
for
the
project.
I
A
J
I
think
it's
very
important
that
that
those
experiments
get
done
rather
than
just
looking
at
percent
inhibition.
At
this
point,
and
obviously
again,
if
you
you
know,
ultimately,
if
you
want
to
do
anything
about
a
little
bit
of
sar
or
something
like
that,
the
percent
ambition
is
is,
is
not
of
that.
It's
interesting
it.
J
H
Yeah,
if
for,
if
we're
just
going
to
target
the
ones
which
we
think
is
john
and
hepatitis,
then
yeah
those
slopes
are
definitely
fine.
Obviously,
if
they're
trying
to
do
all
78
against
both
enzymes,
it
would
take
quite
a
while.
J
No
reason
there's
no
reason,
there's
no
reason
to
do
all
78.
I
think
it's
in
if
you're
just
you
know,
the
goal
of
the
team
is
to
look
at
dual
inhibition.
You
know
just
looking
at
the
ones
or
dual
inhibitors
and
really
profiling.
Those
to
see
again,
I'm
just
very
you
know
these
fragments
in
this
week
activity
you
don't
want
to
be
chasing
ghosts
here,
yeah
and
also
in
terms
of
competition.
I
mean
I
guess
this
is
in
a
level.
J
E
These
would
either
be
the
atp
or
the
substrate
binding
site,
and
let
me
just
check
because
it'll.
A
C
J
It's
on
the
it's
on
the
website,
but
again
I
just
I
I
think
and
allosteric
inhibition
is
obviously
you
know
it's
kind
of
in
vogue
and
somewhat
in
some
disease
areas.
Right
now,
but
again,
as
I
think,
as
was
pointed
out,
you
don't
necess.
I
guess
these
are.
These
are
functional.
Is
that
correct
in
the
sense
that,
in
your
assay
you're,
looking
at
turnover
of
substrate?
H
E
Okay,
yeah
I've
just
got
the
structure
up
here,
but
anyway,
so
the
fragments
are
in
green
and
then
the
purple
is
the
substrate
site.
So
up
here,
so
I
thought
that
might
be
helpful,
but
yeah
it's
on
github.
A
All
right
great,
so
the
action
there
is
to
look
at
the
the
dose
response
and
maybe
to
see
if
we
can
figure
out,
if
they're
being
bound
in
the
same
place.
If
we
know
the
fragment
that
these
come
from,
we
could,
of
course
you
know,
try
to
see
if
they're
binding
in
the
same
place,
I'm
assuming
they
would
be,
but
you
never
know
and
to
do
a
kind
of
cluster
analysis
of
the
actors
versus
inactives.
Those
are
all
the
actions
on
us.
C
Yeah
with
your
with
the
hill
slope,
I'll
go
and
talk
to
adrian,
but
you
know
we
don't
do
we
want
a
full
10
10
point
data
set
for
you
know
each
of
these,
or
can
we
get?
What
would
you
be
happy
for
us
to
get
away
with
concentration?
Well.
J
J
J
You
know
it's
interesting,
that
these
are
functional
in
that
sense,
that
you
know
you're
you're,
you're,
inhibiting
you
know
the
actual
reaction
and
non-competitive,
and
maybe
that's
I
guess
the
question
there
becomes
when
you
start
looking
at
that
binding
site,
how
conserved
is
it
across
the
relevant,
for
example,
escape
pathogens
right
I
mean
one
of
the
things
right
now,
you're
looking
at
you
know
like,
for
example,
muir
d
in
your
d
you're,
looking
at
that
galactic
or
whatever
it's
called
and
we're
not
looking
at
e
coli,
and
so
then,
what's
the
actual,
once
you
start
getting
away
from
the
binding
site,
you
know.
J
What's
the
conservation?
What's
the
conservation
in
that
site?
Three,
that's
allosteric!
Some
of
those
are
kind
of
questions.
I
would
be
asking
at
this
point
in
terms
of
pursuing
that
binding
site
and
getting
a
little
bit
better.
Understanding
of
you
know
if
you're
dual
targeting,
because
that's
going
to
if
you
go
to
a
a
grain
application.
J
Those
are
the
kind
of
questions
your
viewers
want
to
be
asking
right,
they're
going
to
be
asking
okay,
what
you're
deal
targeting
you
know
they
want
to
see.
You
know
the
sequence
they
want
to
see
the
structure
they
want
to
see.
You
know
that,
there's
you
know
a
logical
path
forward
in
terms
of
dual
inhibition
of
relevant
clinical
pathogens.
J
A
B
B
Yeah,
I
guess
I
should
say
on
on
there's
an
action
item
for
me
there
about
going
through
the
literature.
I
thought
that
they
came
up
from
making
making
a
machine
learning
model.
I
think
the
the
bottom
line
is,
if
I
get
the
data
I'm
happy
to
do
it,
I'm
I'm
not
going
to
be
able
to
find
the
time
to
to
go
fall
to
the
literature
also
just
because
I'm
not
really
equipped,
for
it
sure.
C
I
I
I
sorry
got
caught
in
a
call.
My
mother
phoned
yeah,
I
mean,
rather
than
going
to
hill
slopes
with
all
of
those
those
hits,
and
I
think
my
temptation
joe,
is
to
follow
up
on
your
suggestion
around
priority
pathogens
and
see
if
any
of
those
hit
in
dna
of
other
pathogens
as
well.
C
And
then,
if
some
of
those
hit
then
in
my
mind,
may
be
the
ones
to
focus
on
because
you're
right,
we
could
end
up
with
you
know
some
nice
inhibitors
of
group,
c
streptococcus
and
no
one's
gonna
be
interested
in
funding.
That.
J
So,
what's
the
status
with
actually
mere
c
in
your
lab,
because
I
think
he
said
way
back
in
the
day
he
did
life
arc
screen.
What
I
think
was
pseudomonas.
Maybe
have
you
gotten
that
out
I
mean
that's,
that's
not
I
mean
that's.
You
guys
are
driving
the
driving
this.
You
know
project
and
want
to
make
funding
in
the
uk,
but
so
you
know
you
guys
can
decide
which,
with
pathogens
but
again,
I
think
trying
to
zero
in
on
especially
like
again
for
e
coli.
I
would
be
thinking
I
mean
from
your
d.
J
E
J
Assays
up
and
going
and
ultimately
that's
the
data
I
think
you'd
want
to
present
in
the
grant
application
yeah.
C
A
That
okay,
I
yeah
like
I
said
I
don't
really
want
to
go
through
everything
here
and-
and
I
hear
your
comments
earlier
yeah
I
mean
I
guess:
we've
got
a
new
threat
to
everything.
Now
sorry,
I
might
just
leave
that
there
for
the
moment,
but
if,
if
it
gets
really
mothballed,
then
then
we
can
take
it
out
and
I
completely
understood
your
comment
about
bandwidth.
J
Else
here,
so
I
guess
matt,
I
guess
I'm
sorry,
you
kind
of
you
solicited
questions
or
comments.
Yeah,
it's
great
to
see
the
update
on
the
chemistry,
yeah
and
55
mix
of
looks
like
one
of
the
compounds.
I
think.
J
So
I
don't
know
if
you
have
peter
peter
you
have
you
guys
shared
your
your
shipping
address
or
I
guess
it
goes
to
bart
first
or
something
like
that,
but
the
logistics
of
that.
I
guess
it
was
an
action
item
yeah
how
to
send
send
that
compound
on
yeah,
because.
J
Like
if
we
can
get
something
in
you
know
that
compound,
I
guess
we
need
to
add
in
ideally
e
coli
or
acetobacter.
I
guess.
J
J
Yeah,
so
that's
that's
great,
I
mean
if
we
could,
if,
if
they
could,
you
know
if
you
can
get
a
structure,
one
of
that
compound
and
one
of
those
mere
d's,
then
that
would
really,
I
think,
really
form
a
foundation
of
of
a
grain
application
for
you
guys
for
that
in
the
uk.
Folks.
A
Okay,
you
know
what
I'm
gonna,
so
I'm
just
gonna
when
we
have
this
thing
confirmed
and
when
we're
figuring
out
what
we're
gonna
do
in
terms
of
we're
gonna
send
it
and
why
I'm
going
to
ccu
and
joe
to
make
sure
I
get
my
latin
terms
right
and
we're
talking
about
the
right
enzymes.
Okay,
so
to
make
sure
we've
got
a
consensus
on
what
we're
going
to
target
it
against
the
the
you
hangs.
A
J
A
Pretty
paper
local
yeah,
that
no,
that
that's
an
internal
thing,
I
think
you're
gonna,
have
to
start
using
your
own
codes,
because
that's
gonna
confuse
people
so
yeah,
that's
just
in
the
in
the
synthetic
scheme
because
we're
going
from
one
through
to
the
end
one
question
we
had.
We
were
talking
about
this
and
we
wondered
if
there's
any
use
in
having
a
negative
control.
So
a
compound
from
this
series
that
doesn't
have
any
activity
is
that
useful
or
would
that
be
not
useful?.
J
J
I
guess
the
only
comment
on
that
would
be
is,
if
you
well,
first
of
all
you
don't
have
I
mean
you,
don't
even
have
the
assays
to
say
cross-reference.
What's
the
a-z
paper
right,
so
they
z
was
doing
pseudomonas
smear
c.
So
until
sorry,
chris,
but
until
chris's
lab,
has
pseudomonas
mercy
by
a
chemical.
I
say
up
and
running,
and
once
you
had
that,
then
you
could.
You
could
do.
Basically,
you
know
a
couple
data
points
and
then
you
would
have
you
could
use
all
the
az
published.
J
You
know
activity
which
I
think
there's
probably
some
compounds
in
there
that
are
not
as
active.
So
you
could
use
that
as
your
trainings,
you
you
make
that
accessible
and
you
kind
of
have
you
can
combine
those.
Those
two
data
sets
whether
it's
like
data
out
of
chris's,
lab
or
or
the
azp
paper
right.
But
anyway,
it's
again
the
the
critical
step
here
is
going
to
be
at
some
point.
You've
got
to
get
a,
I
guess
a
biochemical
essay.
I
think,
because
right
now,
chris's
lab.
C
A
A
C
C
In
discussion
they're
wanting
data,
so
what
they're
wanting
is
to
have
already
run
the
high
throughput
assay,
so
not
the
assay
becker's
been
doing,
but
the
next
generation,
which
is
very
low
volume,
fluorescent
format.
So
what
I
have
to
balance
is
keeping
elf
happy
and
doing
that
and
keeping
joe
happy
and
doing
mercy.
So
it's
a
very
I'm.
I'm
tall.
G
Just
just
please
bear
me
for
one
last
question.
I
I
think
I
mean
I've
been
searching
for
literatures
for
quite
a
while
about
that,
like
it's
about
mercy
and
murder,
those
of
those
crystals,
the
crystals
of
the
protein
they
have
about
the
crystallographer
of
those
I
I
saw
them
having
like
mandolins
or
magnesium
metals
in
the
center,
but
I
I'm
not
quite
sure
if
they
were
like
originally
endless.
G
Like
the
mercy,
for
example,
in
the
bacteria,
or
do
we
just
add
those
metal
ions
in
so
that
that
might
influence
the
that
might
influence
the
outcome
of
docking
because,
like
when
we,
when
we
are
actually
doing
the
talking,
we
consider
we
consider
about
the
metallic
interactions
or
that
that
depends.
So,
if
that
matters
not
involved
in
the
biological
process,
then
we
probably
might
not
might
need
to
just
delete
that
metal
center
and
just
consider
it
all.
J
G
A
C
D
C
G
I
F
A
F
Mean
sure
sure
I
mean
it's
it
we
can't
tell
what's
water
and
what's
magnesium
is
what
I'm
trying
to
get
it?
It's
the
oh.
You
can
only
tell
from
the
the
environment.
So
if
it's
coordinated
as
a
magnesium
and
it's
a
magnesium,
if
it's
coordinated
as
a
water,
it's
a
water.
We
can't
tell
the
difference
between
the
two
there's,
just
no
physical
way
of
telling
crystallography
of
whether
something
is
a
magnesium
or
a
water.
Every
other
thing,
every
other
element
should
be
more
lucky
because
they
may
have
an
anomalous
signal
about
magnesium.
F
J
B
And
so
on,
I
think
we
do
the
same.
I'd
have
to
just
confer
with
with
casper,
but
I'm
pretty
sure
yeah
that
those
are
removed
just
because
they're,
you
know
it's
uncertain
what
they
are
and
they
can
they're
very
labeled
right.
So
maybe
one
ligand
will
throw
it
out.
Another
one
will
need
it
and
yeah.
So
you
just
end
up,
throwing
it
out
everywhere
for.
J
I
just
you
know
just
a
comment
I
mean
here
I
mean
you
know,
there's
so
much
work
done
in
in
kinases.
You
know,
which
are
obviously
atp
dependent
and
all
that
stuff,
and
you
know
all
these
improved
drugs
and
they're
all
mimicking.
You
know
or
displacing
metals,
and
you
know
the
phosphates
and
all
that
kind
of
stuff.
So
I
mean
it's,
it's
all
around.
You
know
enzyme
concentrations,
and
you
know
it
turns
icy
50s
and
all
that
kind
of
stuff,
but
just
in
terms
of
docking.
Yes,
all
this.
A
All
right,
thank
you.
So
much
great,
I'm
gonna
have
to
run
the
next
meeting.
We
will
technically
be
beginning
of
august.
People
will
probably
be
on
holiday.
It
would
be
nice
to
catch
up
if
people
are
available,
no
pressure.
I
might
just
put
it
in
the
diary.
If
that's
all
right
and
it's
either
the
third
or
the
tenth
of
august,
don't
mind
any
any
immediate
preferences.