►
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/70
On the call: Professor Matthew Todd, Dr Edwin Tse, Yuhang Wang, Kato Leonard (UCL), Laura Diaz Saez (University of Warwick), Lizbe Koekemoer (Uni Oxford/Diamond), Dr Lori Ferrins, Dr Joe Eyermann (Northeastern University), Dr Jan Jensen (University of Copenhagen), Dr Chris Swain (Cambridge MedChem Consulting), Dr Jan Abendroth, Dr Bart Staker (SSGCID).
A
All
right,
hi
everybody,
it's
tuesday,
8th
of
march.
This
is
osa
merly
gays
meeting
and
I
will
share
screen
so
that
you
are
seeing
what
I'm
seeing,
which
I
think
is
probably
this
guy
and
I
just
need
to
move
that
out
of
the
way.
So
you've
got
that:
okay,
okay,
so.
A
Just
want
to
make
sure
we've
got
the
right
people-
okay,
so
I
guess
the
the
last
time
we
were
looking
at
a
bunch
of
data
on
the
from
becca
about
the
dual
inhibition
capabilities
of
these
molecules
that
dana
had
synthesized
and
saw
the
data
that
suggested
that
some
molecules
appear
to
be
giving
some
dual
inhibition
where
the
correct
dose
response
curve.
A
We
had
taken
one
of
those
compounds
and
we
have
designed
some
variants
of
that
which
seem
reasonable
in
terms
of
exploring
the
compound
that
was
looking
the
most
reasonable.
I
think
they
are
if
I've
linked
this
correctly
here
so
cato
and
ed
have
been
playing
around
with
the
structures
which
gave
the
nice
results
just
so
we
can
have
a
go
at
some
more
compounds
for
enzymatic
screening
and
I
think
these
all
look
reasonable.
A
They've
got
you
know
extra
mass,
varied
possibilities
for
evaluation,
they've
they're,
reasonably
inexpensive
they've
got
good,
calculated
log
p
values.
So
unless
there's
any
cries
of
dissent,
we're
going
to
go
ahead
with
making
a
bunch
of
those
and
sending
those
off
to
warwick
in
the
short
term.
A
I
wasn't
sure
what
the
action
was
there
for
either
laura
or
becker
to
maybe
maybe
comment
on
in
terms
of
working
out
what
these
compounds
might
be
doing.
Does
anyone
want
to
address
that
now.
B
So
hi
I'm
laura,
I
think
I
was
gonna,
do
some
essays
for
competition.
I
can
say
tp.
I
didn't
see
her
on
the
list
of
people
joined,
but
I
know
she
started
doing
the
essays
so
maybe
later.
C
B
She
joins,
then.
Maybe
she
can
address
that
and
but
I
haven't
seen
her
so
haven't,
been
able
to
talk
to
her.
A
A
And
and
then
on
the
these
compounds
I
mean
you
know:
we've
measured
their
ability
to
inhibit
dmr
e.
I
guess
that
we
were.
We
saw
your
data
last
time,
laura
of
the
crystallography,
that
you
were
doing
with
these
compounds
and
proteins.
Yes,
and
you
were,
I
think
you
were
going
to
try
and
pursue
co-crystallization
experiments.
Do
you
have
enough
that
you
want
to
share
or.
B
I
haven't
gotten
good
crystals,
yet
there
are
some
crystals
growing
but
they're
very
tiny.
So
I'm
trying
to
optimize
those
and
see
if
I
can
get
a
bigger
crystals.
They
are
micro,
crystals,
basically,
and
then,
if
that
progressed
then
obviously
I
will
be
doing
testing
of
the
fraction
and
collection
in
april
because
the
diamond
is
shut
down
until
april
the
first
week
of
april,
so
a
few
more
weeks
and
see
if
I
can
get
something
in
a
few
more
weeks.
A
D
E
But
he's
behind,
he
can't
look
at
his
computer
and
his
stuff
right
he's
not.
A
Okay,
all
good-
we
can
do
that
later.
In
that
case,
that's
fine,
okay,
so
I
and
we
we
heard
last
time-
data
from
becca
and
dave
from
chris
thompson's
team
on
the
enzymatic
essays.
If
I
could
just
try
and
encourage
the
team
to
post
the
data
that
they
had
because
it
was,
we
saw
it
on
the
screen
as
part
of
the
presentation,
but
it's
always
nice
to
have
the
data.
If
we
can
for
this
project
post
it
somewhere.
B
A
All
right
thanks
because
yeah
I
mean
it
started
a
conversation
last
time
about
how
these
inhibitors
are
performing
in
the
sense
that
we,
you
know,
we
saw
inhibition
with
some
compounds
that
were
known
from
the
asha
zenica
work
to
be
highly
potent
inhibitors,
but
we
were
getting
thousand-fold
differences
from
those
potencies
in
the
assay.
So
I
guess
it
would
just
be
useful
to
have
the
data
there.
So
we
can
see
exactly
what
was
done.
B
D
D
Thousands
group
on
you
know
your
ligase
enzymology
and
stuff,
and
so,
but
I
know
they
were,
they
were
going
to
look
at
things,
but
I
think
in
general
the
the
relative
potencies
are
okay,
I
think
the
the
issue,
the
concern
I
had
was
just
being
able
to
identify,
really
weak
inhibitors,
and
I
think
that
was
something
they
were
gonna
look
into,
but
yeah,
I
guess
chris
is
not
actually
excuse.
Me
adrian
is
not
on
the
line
this
time,
but
we
can
follow
up
on.
I
guess
next
time.
A
B
D
B
D
B
A
D
D
There
are
very
few
examples
of
actual
crystal
structures
of
an
atp
bound,
mure,
ligase
inhibitor
other
than
the
az
compound
I
mean
everything
else
is
like
is
is
basically
adp
or
you
know
there
was
stuff
that
was
done
in
the
ema
pocket,
but
in
terms
of
atp
itself,
the
only
really
good
structures
we
have
well
actually
there's
the
second
one,
which
is
the
diamond,
miri
small
fragment
but
yeah.
There's.
C
A
Okay,
so
the
the
other
things
I
guess
is
talking
about
the
the
enzymatic
assay
is
that
we
obviously
were
hoping
to
get
evaluation
of
the
atomized
compounds
in
that
assay.
Do
you
happen
to
know
or
if
that's
been
taking
place,.
B
So
I
know
that
we're
trying
to
address
the
problems
with
them,
okay,
eyes
and
seeing
if
they
can
lower
down
the
concentration
of
enzyme
to
the
high
throughput
screens.
So
the
last
time
it
was
working
and
they
were
gonna
start
planning
the
screen.
But
I
don't
know
if
you
know
everything
went
okay
in
the
last
checks
or
not,
because
adrian
is
away
as
well.
So
I
haven't
been
able
to
update
with
anyone
this
week
yeah.
B
A
All
right,
great
so,
just
before
we
talk
about
anything
on
the
spr
just
wanted
to
mention
this
thing
that
I've
been
in
email
correspondence
with
with
chris
and
laura
and
others
about,
which
is
this
rather
rather
cryptic
thing.
A
So
we
got
some
money
from
epsc
and
ucl
to
run
a
competition
over
the
coming
four
months,
basically
until
the
end
of
june
to
to
run
a
competition
to
ask
people
working
in
the
ai
machine
learning
field
who
are
using
generative
methods,
so
not
virtual
screening
of
large
libraries,
but
generating
molecules
that
that
meet
certain
criteria
that
will
bind
to
a
merligase.
A
So
the
idea
is
that
when
the
competition
is
started
which
we
wanted
to
do
to
do
this
week,
we
will
be
going
out
to
a
bunch
of
companies
and
academic
groups.
Saying
do
you
want
to
have
a
go
at
this
challenge?
If
you
can
suggest
molecules
for
us
to
purchase
or
make
we'll
do
that
and
we'll
send
them
to
warwick
for
evaluation
and
we'll
try
and
fit
in
as
many
cycles
of
that
as
we
can
before
the
end
of
the
grant?
It's
very
short
term.
A
This
is
going
to
be
done
pretty
quickly
and
ed
will
be
doing
the
synthesis
and
the
procurement
of
the
molecules.
You
know
we've
done
this
before,
with
with
open
source
malaria
for
a
phenotypic
drug
discovery
project.
This
time
it's
target
based,
and
it's
specifically
about
generative
modelling,
which
is
which
is
tough.
A
The
key
is
to
make
sure
that
the
rules
are
simple,
so
the
operational
rules
are
simple.
We
know
how
to
run
this
kind
of
thing,
but
the
science
rules
need
to
be
really
crystal
clear
for
this,
and
I
want
to
make
sure
that
we're
giving
people
the
right
advice,
so
I
want
to
make
sure
we're
giving
them
exactly
the
right
target
structure
to
work
with
and
any
clues
we
might
have
as
to
other
compounds
like
fragments
that
are
binding
in
the
same
place
that
we're
interested
in.
A
F
I
I
guess
I
don't,
but
I
have
a
question
on
this.
I
mean.
Is
this
the
is
this
this?
We
we
dock
to
a
different
target
right,
a
different,
crystal
structure,
but
is
this
more
d.
A
Or
more
c,
I
guess
it
doesn't.
It
can
be
anything
I
want
it
to
be
something
that
we
think
will
will
give
us
the
that
we
can
measure
right,
firstly
and
secondly,
that
will
be
of
maximal
use
according
to
b,
according
to
people's
views
on
this
call.
A
Yeah
so
you're
it
depends.
I
I
didn't
clock
whether
your
method
was
was
generative
or
whether
it
was
linked
to
specifically
virtual
screening
of
commercial
libraries.
F
Oh
no,
no
it's
I
mean
no,
it's
generative,
so
the
models
I
mean
with
one
exception.
The
models
we
predict
are
completely
normal
right
can't
be
bought.
I
mean
in
the
sense.
G
A
Well,
it
depends,
I
mean
it
depends
how
complicated
the
molecules
are
right.
So,
on
a
related
project,
we
were
working
with
some
people
on
a
degenerative
model
for
for
the
coronavirus,
nsp,
13
helicase,
and
I
mean
of
about
30
molecules
or
so
came
in
from
that
within
certain
constraints
of
the
pharmaceutical
model,
molecular
weight
and
log
p
and
everything
and
half
of
them,
roughly
half
of
them
are
purchasable.
A
Yeah,
I
don't
so
it
seems
like
you
know
if
we
have
a
few
fragments
bound
somewhere.
That
gives
people
a
starting
point,
but
the
guy
I
was
talking
to
at
ucl
who
is
going
to
be
entering
this
thing
and
getting
people
of
you
know
interested
said
he
doesn't
need
anything
bad.
He
just
needs
the
protein
structure.
Okay,.
A
D
D
So
I
guess
the
one
thing
my
input
to
you
matt
would
be
that
you
know
you
need
to
kind
of
work
through
with
obviously
what
chris
thousands
group
actually
which
ice
design
that
they
feel
they
can
actually
run
the
assay
on
again.
That's
that's
been
yeah.
I
guess
that's,
that's
the
key
thing
right.
Ultimately,
you
want
to
test
these
biochemically
and
you
know
which
ashley
they
feel
they
have
the
most
confidence,
comfort
level
ability
to
screen
compounds.
D
D
Data,
we
haven't
seen
very
much
data,
so
you
know
the
practical
point
here
is:
if
you're
asking
people
to
you
know
put
effort
in
to
you
know,
generate
ideas,
it
would
be
great
to
see
them
tested.
You
know,
I'm
waiting.
I've
been
waiting
three
years
now
for
my
compounds
to
be
tested.
So
that's
why
I'm
a
little
yeah.
I
know
I
understand
that's
sensitive
about
this
issue
about
actually
getting
data
out
of
the
yeah.
B
B
Yeah,
I
understand
you
we
also,
so
I
also
offer
spr,
which
is
where
I
can.
You
know
I
can
be
of
a
better
input
because
I'm
not
doing
the
essays
yeah
so
spr.
I
can
get
that
done
pretty
fast.
A
A
Yeah,
so
in
that
case,
if
we
wanted
to
screen
these
compounds
by
spr
for
your
from
your
perspective
and
if
we're
talking
about
you
know,
I
don't
know
30
compounds
a
month
or
something
something
like
that,
something
like
that.
What
should
we
ask
people
to
predict
binders
for.
B
So
things
like:
what
do
we
want
to
use
these
compounds
for,
if
it's
again
for
dual
inhibitors
or
multiple
inhibitors,
then
I
think
we
should
keep
going
with
the
same
pockets
that
we're
doing
the
atp
side,
for
example,
or
something
like
that,
but
from.
B
Not
yeah
for
mere
d
or
me
or
e
things
that
we
are
already
working
on
right,
build
up
or
continue
building
up
on
the
work
that
we
are
doing.
I
don't
think
we
should
start
something
completely
different,
but
in.
A
B
A
And-
and
that's
the
most
so
in
terms
of
the
spr
assay,
then
molecules
that
are
predicted
to
bind
that
structure
should
come
up
should
come
up.
Ideally
on
the
spr
essay.
B
You
should,
I
can
also
run
in
the
presence
of
substrates
or
in
the
presence
of
all
the
molecules,
and
we
can
have
a
semicomplex
yeah.
I
can
do
a
few
things
on
that,
but
yeah
on
the
the
spire
in
in
principle
is
an
apo
protein
and
right,
but
then
I
can,
I
could
add
more
things
into
the
buffer
there's
a
buffer
if
we
want
to
have
different
complexes
as
well
as
a
compound.
B
A
B
A
Perfect,
I
mean,
I
think
we
have
to
go
with
that.
We
we
originally
planned
this
project
to
be
one
versus
the
nsp
13
helicopters,
but
we
can't
we
can't.
We
can
only
access
an
atpa's
asset.
We
can't
access
a
bike
by
physical
assay.
A
D
A
So
they,
the
the
second
I've
got
this
here,
don't
need
to
talk
about
it.
A
You'll
be
pleased
to
know,
atomizers
have
reached
out
and
we
are
due
to
set
up
a
call
with
them.
The
reason
why
our
original
inquiries
were
going
unanswered
is
because
our
original
sign
contact
left
atomize
a
year
ago,
but
but
no
one
had
an
out
of
office
on
so
we
we
now,
I
think,
have
a
new
person.
I
think,
and
we
will
be
in
touch
with
them
asap.
A
So
this
was
the
I
mean
it's
in
the
original
proposal
document,
the
one
that
chris
swain
actually
submitted
and
their
determination
was
from
the
three
uag
structure.
These
were
the
relevant
residues
and
yeah.
I
was.
I
was
weighing
on
just
some
insights
from
from
the
from
the
warwick
team
about
whether
or
not
the
compounds
seen
on
the
crystal
structures
were
interacting.
A
B
Oh
yeah,
there,
you
are
sorry
I
was.
I
just
thought
it
was
yours
in
your
email
because
it
was
smart
yeah.
So
there
were
some
recipes
that
were
in
common
between
both,
but
they
were
not
key
interactors
with
any
of
the
hits
that
we
got.
D
I'm
sorry
I'm
so
confused.
So
so
we
have
the
what
you
know
the
x-chem
fragments
right
and
in
this
quote
what
we
think
we
might
be
calling
an
alistair
pocket,
correct,
yeah
and
we
have
the
at
poc
atp,
pocket
and
or
adp
pocket,
and
we
have
uma
pocket
right
yeah.
So
these
compounds
from
atomwise
are
they
in
the
atp
pocket
or
there's.
B
B
B
All
right,
so
we
have
done.
We
have
done
two
runs
of
xcom,
okay
right,
one
with
the
fragment
screening
from
the
x
game
library,
and
then
I
did
the
otherwise
library
and
so
on
the
first
screen.
We
got
different
binders
and
one
some
of
them
were
binding
on
the
allostatic
site
or
what
we
are
calling
analystic
site
and
then
in
the
case
of
the
otherwise
all
of
the
hips
were
in
the
anaesthetic
side.
A
D
I
I
think
I
would
you
know,
because
you
have
data,
I
I
would
suggest,
maybe
going
actually,
you
know
driving
it
on
the
allosteric
site,
that
pocket
and
focusing
focus
in
on
that
and
let
people
generate
and
that
kind
of
they
have
some
data
that
you
know
you
can
obviously
you're
sharing
publicly,
and
then
you
know
that
just
gives
you
some
more
info
on.
You
know
supporting.
You
know
that
that
area
of
design,
hypothesis.
D
Struggling
you're
struggling
you're
start
we're
we're
struggling
to
get.
You
know
in
your
d,
we're
struggling
to
get
any
kind
of
we
don't
like,
so
we
don't
really
have
a
good
reference
other
than
you
know,
adp
or
adp,
whatever
pcp.
Whatever.
That
thing
is
so
just
in
terms
of
having
data
to
kind
of
support,
all
the
ideas
people
are
going
to
submit.
I
would
think
that
was
their
pocket
would
be
the
one
that
you're
going
to
have
more
resources
to
evaluate
the
the
submitted
ideas.
A
Yeah,
so
so
some
of
the
people
interested
in
this
would,
as
you
just
said,
like
any
information
we
have
about
small
molecules
that
bind
in
the
same
place
sounds
like
some
don't,
but
some
do
so.
Presumably
what
we
can
do
is
say
well,
this
is
the
thing
that's
on
the
screen
here
was
the
pocket
we
defined
for
the
atomized
composition.
Presumably
a
subset
of
those
residues
are
the
ones
that
are
in
touch
with
the
frac,
with
some
fragments
that
we
got
from
the
screen.
The
x-cam
screen,
I'm
assuming,
because
it's
the
allosteric
pocket
right.
B
Yeah,
so,
okay,
so
from
all
those
residues,
there
will
be
more
that
will
be
shared
with
the
first
skm
screen
in
terms
of
contact
with
compounds,
because
there
were
more
hits
in
different
areas.
In
terms
of
the
aesthetic
side,
it
will
be
the
same
result
or
very
similar
result
to
the
atom-wise
screen
in
the
x-cam,
because
it's
the
same
pocket.
A
A
Sure
yeah,
so
what
I
was
hoping
to
do
is
share
this
and
then
share
the
the
the
small
number
of
relevant
structures
which
contain
the
fragments
that
were
bound
to
the
same
place.
And
so
basically,
what
we'd
like
is
the
design
of
compounds
that
are
higher
molecular
weight
than
the
fragments
but
which
are
binding
in
the
same
place.
Yep.
A
That's
that
was
the
that
was
kind
of
the
idea,
so
we
could
share.
In
that
case,
we
can
share
all
of
the
structures
which
have
the
fragments
bound
in
the
outer
pocket
and
we
can
share
the
structures
you
found
at
the
atomized
compounds,
also
about
allosteric
pocket
and
say
ultimately
that
we
want
compounds
bound
here,
yep
right,
so
I
guess
what
on
us
is
to
list
all
of
those
those
relevant
structures
so
that
people
know
what
we're
talking
about.
A
Well,
I
mean
best
best
kd
right
or
something
or
yeah.
You
see
that's
a
good
point
right.
So
do
we
go
for
the
best
result
or
the
best
ligand
efficiency?
A
Don't
want
to
make
that's
iterative,
I
mean
we
had
that
in
the
malaria
competition
there
were
some
really
beautiful
structures
that
were
suggested
as
thing
as
targets
for
that
and
we
and
we
basically
had
to
say
we
can't
make
that
there's
just
no
way
in
the
time
frame
we
have
available.
So
we
we
will
be
going
back
to
people
who
would
provide,
elaborates,
really
elaborate
structures.
They
can
simplify,
please
or
make
it
or
yes
or
make
it
yourself.
Yeah
yeah.
A
Okay,
thanks
for
that,
I
just
it
it
may
not
lead
anywhere
or
it
may
lead
to
a
brand
new
set
of
compounds
that
are
useful.
You
know,
ultimately,
if
we
got
a
micro,
molar
binder,
it
would
be
really
fantastic.
So
it's
worth
a
shot
and
we
have
some.
You
know
resources
for
for
four
months
of
work,
so
we
may
as
well
go
for
it.
D
So
so
chris,
I'm
sorry
matt,
I'm
sorry
again,
just
I
don't
know
in
terms
of
the
I
don't
really
call
it
the
parameters
of
what
people
can
actually
do
in
terms
of
this
quote.
We
call
it
or
not
in
order.
D
Generator
excuse
me
generative
methods,
I
mean
so
at
that
level,
can
they
use
one
of
the
existing
fragments
and
then
use
whatever
pharmacoport
touch
points
in
the
binding
site
and
then
kind
of
do
some
kind
of
de
novo
iterative
design
based
on
you
know,
using
that
initial
fragment
as
a
starting
point
is
that
was
that
legit?
I
think
it's
legit.
D
Okay,
I'm
just
thinking
that
you
know
that's
I
mean
I
think
the
more
you
can
define
kind
of
what
approaches
can
be
used
would
be
probably
useful
to
the
whoever's
going
to
be
competing
or
submitting
ideas.
Yeah.
A
So
I
I
was
going
to
put
a
caveat
in
in
the
rules
about
exactly
that.
So
what
I
was
going
to
say
is
that,
if
you're
going
to
use
an
existing
binder
to
predict
a
new
one,
that
the
the
new
compound
needs
to
be
at
least
three
heavy
atoms
heavier
than
the
original,
which
usually
means
you're,
adding
on
about
50
molecular
weight
kind
of
thing.
A
But
but
yeah
we'll
see,
I
mean
that's,
not
that's
kind
of
not
the
idea,
but
but
it's
not
it's
not
it
wouldn't
be.
It
would
still
be
useful
right
if
that
happened,
because,
as
you
know,
anyone
who's
done
fragment
work
right.
Just
elaborating
a
fragment
is
tough.
A
All
right,
all
right
great,
I
think
we
should
move
on
just
in
the
interest
of
time.
Yeah
are
you
now
with
us
at
a
desk,
or
are
you
still
on
the
road?
E
A
Okay,
is
it
a
good
time
to
have
a
conversation
about
how
best
to
get
structures
sure
thing
so
we?
Yes,
you
you
posted
very
nicely
a
new
structure
which
joe
was
looking
at
wait
a
minute.
Let
me
just,
I
think,
I'm
looking
outside
here
and
there.
It
is
so
yeah
new
structure,
which
was
great,
so
pseudomonas
movie
and
complex,
with
adp,
really
fantastic
to
see
that
and
then
joe
was
posting.
A
C
Well,
I
mean
the
approach
of
truncating
proteins
to
get
better
structures
or
get
more
easily
crystals.
That's
a
well-known
approach.
It
has
been
employed
here,
for
instance,
for
any
of
the
mirror
c
proteins
if
we
call
it
anything
other
than
a
b1
or
bb1
in
our
ssh
crd
nomenclature.
C
It's
a
model,
it's
a
truncated
protein.
So
from
you
see,
we
have
quite
significantly
truncated
the
protein
that
eventually
gave
crystals.
I
think,
for
mere
d.
We
had
two
different
constructs.
I
know
for
sure
we
had
two
different
constructs
on
the
in
the
running
and
I
think
for
one
target,
one:
the
full-length
protein
crystallized
better
for
the
other
target,
the
slight
truncated
protein
crystallized
bed.
C
C
But
you
also
run
into
issues
there.
That
proteins
might
not
express
if
you
truncate
a
certain
region,
even
though
it
might
be
disordered
in
the
crystal.
If
we
are
going
for
that,
we
should
consider
timelines.
C
We
wouldn't
be
able
to
get
purified
protein
within,
like
at
the
very
very
best
two
or
three
months.
I
would
say
so.
If
you
want
to
have
something
right
now,
we
are
better
off
trying
to
get
more
protein
or
just
crystallizing
the
same
target
again.
B
So
yes,
also,
these
buildings
are
very
flexible
and
we
are
seeing
some
a
lot
of
movement
on
the
loops
that
might
be,
and
so,
for
example,
for
the
aerostatic
side,
there's
interactions
with
a
loop.
So
if
we
remove
that
loop
because
it's
not
pressing
another
structure
is-
and
it's
made
a
contact
with
a
new
compound-
it's
a
bit-
it's
quite
difficult
to
know
which
parts
to
remove
from
the
structure,
even
if
they
are
not
seen
in
other
questions.
D
That's:
okay!
That's
when
you
guys
are
done
so
matt.
Do
you
have
go
farther
down
where
I
have?
I
just
posted.
I
guess
this
morning
or
yesterday
it
was
the
various
structures
of
the
mirror-like
aces
yeah.
So
so
I
think
you
know
the
the
main
thing
here.
I
guess
is
that
as
here
so
so,
these
are
some
of
the
examples
of
the
difference:
you're
seeing
your
d,
mere
e,
different
species
and
so
on.
D
So
so,
if
you
can
see
the
the
truncation
in
mirror
c,
so
it's
the
the
normal
the
normally
these
things
have
like
400
and
some
amino
acids,
and
so
in
your
c
they
truncated
off
about
100
or
so
amino
acids
and
the
part
that's
truncated.
Is
this
domain
kind
of
on
the
right
hand
side?
Do
you
see
whether
it's
in
rear
d
or
mirror
e?
D
So
this
is
very
flexible
and
I
think
that's
that's
what
that's
guess.
That's
the
my
thinking
about
you
know
you
know.
Do
we
if
we
remove
that
now
and
all
the
issues
that
you
was
talking
about
in
terms
of
you
know,
does
the
proteins
still
express?
Can
you
get
enough
protein?
If
you
make
that
kind
of
a
truncation
or
not
so
there's?
Actually
this
engineering
challenge
and
unknown?
You
know
uncertainty
about.
D
You
know
what
would
actually
happen
for
these
different
species,
different
mirrors
or
ligases
and
so
on,
but
it
just
seems,
like
you
know,
the
there's
just
been
so
many
challenges
I
mean.
Even
you
know
yawn,
and
you
talked
about
you
know
getting
the
structure.
Just
pseudomonas
verde
structure,
the
one
in
the
middle
top
middle
purple
structure
was,
was
ultimately
got
a
nice
data
set,
but
I
mean
it
was
and
you
had
the
actual
april
I
didn't
put
the
apo
in
here.
You
know
oppo,
you
know
there
was
a.
C
D
The
right,
so
it
closes
down
somewhat
on
binding
of
adp.
D
So
anyway,
I
guess
you
know
from
I
guess,
from
a
structure-based
direct
design
viewpoint.
You
know
it
would
be
nice
to
have
a
system
that
you
know
would
really
work
well,
and
I
just
I
guess,
that's
just
my
thought.
You
know
that
is
it
worth
thinking
about,
for
example,
for
example,
mere
d
of
of
trying
a
truncated
version
and
seeing,
if
that
ends
up
behaving
better
from
getting
crystal
structures.
So
that's
a
thought,
but
it's
just
you
know.
D
Clearly
the
there
was
quite
a
bit
of
work
done
ssgcid
on
the
pseudomonas
new
sea
there
were,
there
were
four
different
or.
B
D
There's
actually
five
different
constructs
that
were
generated
to
get
the
structure
for
the
pseudomonas
merci,
so
obviously
you
know
there's
work
involved
and
you
know
it's
a
matter
of
you
know
it's
just
something
that
you
know
the
team
feels
is
worthwhile
doing,
and
I
guess
you
know
I
guess
you
can
kind
of
guess
what
my
view
is
but
yeah.
This
is
there's
a
big.
This
well.
E
No,
I
can't
guess
because
I,
when
I
look
at
this
figure
this
way,
it
looks
to
me
like,
with
that
extra
lobe,
it's
making
ligand
contacts
and
so
yeah.
I.
E
D
Actually,
I
don't,
I
don't
think,
that's
necessary.
So
if
you
look
it's
actually
the
az
compound.
Well,
okay,
we
don't
necessarily
know
I
mean
so.
My
original
sar
comment
about
why
we
weren't
seeing
activity
in
your
d
for
the
az
compound
was
because
of
that
helix.
D
If
you
look
at
the
helix
there
in
the
pseudomonas
verde,
it
looks
like
it's
jammed
up
against
it,
but
we
know,
but
you
know
we
know
that
if,
if
the
lobe
was
on
your
c,
if
you
had
to
fully,
if
you
had
the
full
version
of
your
c,
you
know
it
would
have
that
that
second
lobe
right
that
second
domain
and
that
that
helix
is
you
know
if
you
look
at
all
the
other
structures
like
helix
is
somewhere
in
the
vicinity,
but
we
know
that
those
ac
compounds
are
almost
subname
or
against
mere
c.
D
So
somehow
you
know
you
can
have
that
that
lobe
and
that
helix
of
that
other
domain
basically
stacked
up
against
that
ac
compound
in
a
in
a
positive
way,
but
it
doesn't
seem
to
work
based
on
laura's
data.
I
think
I
think
it
was
laura's
data
or
adrian's
data.
Last
last
meeting
that
they
weren't
really
seeing
any
activity
against
your
d
with
juhan's
compound
the
private
call.
D
So
so
I
I
think
I
think
the
interactions
you
probably
want,
if
you
think
about
at
least
okay,
my
my
my
philosophy
or
strategy
would
be.
You
know
the
when
becca
and
I
kind
of
worked
on
designs
here.
Most
of
the
interactions
are
with
this
conserved
asparagine
and
then
there's
a
helix
on
the
back
side
that
the
phosphates
interact
with,
and
that's
where
you
know
most
of
our
design
strategy
is
going
towards,
which
does
not
involve
any
of
the
residues
kind
of
involved
with
that
that
lobe
on
the
right.
D
So
I
would,
I
would
actually
avoid
that
forgiven.
You
know
for
the
and
you
don't
see
you
know,
I
don't
didn't
do
the
superposition,
but
if
you
superimpose
these
various
mirrors
and
if
you
look
at
kind
of
the
atp
site
itself,
it's
hot,
you
know
it
is
fairly
highly
conserved,
and
so
you
go
after
the
you
know
the
conserved
interactions
within
the
atp
adb
site.
D
Anyway,
I
mean
that's,
I
you
know
again,
that's
you
know
it's
someone
else's
has
to
put
all
the
elbow
grease
in
to
actually
generate
it
and
you
know
generate
the
clones
and
express
them
purify.
Do
fight
crystallography
et
cetera,
et
cetera,
but
it
just
seems
like
we've
struggled
to
get
good
crystal
systems
at
this
point,
and
so
this
would
be
one
one.
Maybe
one
thought
we
could
consider
one
approach.
E
Yeah,
so
we
can
do
that
like
that's,
we
can
put
them
and
serve
in
our
pipeline.
John
is
correct
that
there's
a
bit
of
a
backlog
and
we're
in
the
middle
of
our
competitive
renewal.
E
So
you
know,
I
think
we
have
about
five
more
months
of
funding
left
or
so,
but
we
can
make
those
and
put
them
in
our
in
our
you
can
even
make
them.
Actually,
if
you
want
to
design
constructs
and
send
them
to
me,
I
can.
I
can
put
them
in
for
sure.
B
B
Me
is
that
we
will
be
missing.
I
understand
your
point
joe,
but
I
was
like
every
time
I'm
thinking
about
that.
It
just
makes
me
worried
that
we
are
missing
as
well
all
their
contacts
but
but
yeah.
But
it's
true
that
we
are
struggling
in
general
to
get
a
good
system
for
crystallography
so
far,
but.
C
B
B
So
the
microcrystals
is
the
clock
crystallization.
So
I
do
get
april
crystals
for
mu
e
and
mu
d.
In
the
case
of
mu
d,
those
are
the
morphism
ones
and
I
need
to
test
it
against
more
compounds,
but
from
the
first
screen,
a
small
thing
that
I
did
there
was
only
one
hit
and
it
was
in
this
pocket
in
a
crystallization
contact
and
it's
really
nice
pocket
for
things
to
be
binding
to
so
it's.
C
C
B
C
And
that's
the
story.
We
run
into
various
branches
of
operation
several
times
that
early
on
in
a
program
where
we,
when
we
have
weak
compounds,
we
can't
really
stabilize
our
wanted
confirmation
good
enough
that
we
can
reproduce
that
in
crystallography
and
we
we've
been
trying
and
trying
and
trying
and
eventually
when
chemistry
provides
good
enough
compounds
that
really
stabilize
the
confirmation
that
we
intend.
C
Then
things
work
much
easier.
But
it's
hard
for
us
in
crystallography
to
to
sort
of
anticipate
that.
A
Sorry,
some
of
you
may
have
already
said
this,
but
but
having
a
doing
the
full
length
mercy
rather
than
rather
than
a
truncated
murder,
was
that
something
which
would
also
take
time.
This
is
a
new
request
right.
D
D
You
make
yeah,
so
they
were
as
as,
as
said,
matt,
you
know.
So,
basically
they
did
four
different
constructs
at
least
four
and
the
several
of
them
were
full
length
and
they
clearly
either
they
didn't
get
a
good
expression
enough
material
or
they
didn't
get
good
crystals
so
that
they
ultimately
the
only
way
they
got
a
structure
with
ac
compound
was.
Was
this
truncated
version.
D
D
I
I
put
up
the
I
think
you
scroll
back
up,
yeah
matt,
I
have
those
constructs
was
in
the
right.
There.
C
E
And
I,
and
we
all
of
the
purifications
failed
for
like
one
and
three
trunk,
the
full
length
and
the
one
and
truncation
one
and
truncation
2.
When
I
look
at
those
in
our
database
as
well
as
truncation
5.
Anything,
that's
basically
larger
didn't
make
it
so
it
never
actually
made
it
to
yawn.
To
attempt
when
I
look
at
it.
E
We
we
pretty
much
as
our
standard
procedure
is
to
send
the
full
length
through
first
whenever
we
do
anything
in
the
target,
so
we
would
have
done
all
the
full
lengths
before
we
did.
The
truncations
doesn't
mean
they
couldn't
be
traded
again
and
also.
C
E
E
Fails
and
we
move
along
to
the
next
thing,
so
there's
there's
room
for
more
work
here
and
we
do
have
the
materials
you
know
we
have
the
clones
and
stuff
laura.
Did
you
ever
receive
your
stuff
from
us.
B
B
Oh
yes,
yeah
excuse.
C
C
B
B
B
C
B
Also
waiting
for
the
e-coli
constructs
because
I
also
requested
e
coli
constructs
and
they
are
being
cloned,
so
I
was
waiting
for
it
to
arrive
and
then
do
all
the
expressions,
purifications,
etc.
A
B
Yeah,
so
I
could
follow
up
after
the
cloning.
You
know
if,
if,
for
example,
ssccid
can
do
the
cloning
I'm
happy
to
take
the
expression
qualification,
maybe
that
way
is
faster.
I
don't
know,
I
don't
know
how
long
it
will
take
in
the
cloning
pipeline.
Basically,.
A
Okay,
just
in
the
interest
of
time
wanted
to
make
sure
we
cover
anything
else
that
we
need
to
do
here.
So
you
you,
hang
I'm
gonna
sort
of
summarize.
For
you,
sorry,
you've
posted
a
your.
Your
synthesis
of
the
a
z5595
compound
is
done
and
you're
now
making
steady
progress
towards
the
amino
version
of
that
compound,
which
is
great
so
so
work
is
ongoing
there
and
your
hands
making
steady
progress
for
the
combination.
Modeling
yan
already
posted
some
of
that
instructors.
A
Last
time
for
the
last
meeting,
we've
just
done
a
synthesized
ability
and
procurement
analysis
here,
which
I
just
need
to
run
through
with
you
hang
on
with
ed,
so
that
we
can
think
about
which
compounds
we
can
get
still
and
which
compounds
we
might
want
to
make.
Given
that
there
are
some
big
disruptions
at
the
moment
to
synthesis
on
demand,
so
we
just
need
to
have
a
fresh
look
at
how
we
get
some
of
those
compounds
and
evaluate
them.
A
And
there
was
oh
yeah
laurie
found
this
merc
open
global
health
library,
a
bunch
of
compounds
available
from
merck
that
we
could
request
for
free
if
we
wanted
to
they
come
in.
This
forms
are
30
microliters
and
a
constitutional
10
millimolar
in
dmso.
We
can
just
acquire
those
compounds
if
we
want
to
and
we
will
be
able
to
post
the
data
on
any
hits
we
get.
I
think
so.
A
I
was
hoping
to
convince
the
warwick
team
to
act
as
receivers
of
the
compounds,
which
means
that
you
would
have
to
engage
with
merck
in
terms
of
signing
a
piece
of
paper,
so
I
can
bring
that
up.
I
was
having
to
bring
it
up
with
chris
today,
but
he's
not
around
so
I
might
do
that
offline,
but
that
could
be
another,
a
bunch
of
compounds
that
we
may
not
have.
I
don't
know
what
the
library
contains
actually,
but
it
could
be
a
different
set
of
comments.
The
ones
we've
already
evaluated,
potentially.
A
Okay,
okay
and
then
there
are
some
other
multiples
there
and
stuff
all
right
did.
I
forget
anything
or
anyone
wants
to
raise
anything,
just
a
quick
question
on
this
competition.
When,
where
would
you
announce
it
well
here?
Actually,
okay,
wait
a
minute
anyways,
it's
it's
draft
still,
but
it's
it's
that
sorry
we're
just
not
quite
not
quite.
A
But
we'll
we'll
once
we
once
we're
happy
with
the
text
and
now
that
I
know
that
we're
happy
with
the
target.
We
can
tweak
this
and
and
get
going.
There's
a
guy
called
brooks
page
at
ucl.
He's
part
of
the
turing
institute
who's,
one
of
the
the
machine
learning
guys
who's
driving
it
and
peter
kovany
does
a
bunch
of
stuff
on
predicting
molecules
of
binding
proteins
in
chemistry
here
at
ucl,
and
we
got
this
money
to
run
a
composition,
but
we're
going
to
try
and
involve
others.
A
The
aim
of
the
competition
is
to
develop
a
community
of
generative
people
around
a
particular
problem
and
share
what
works
well
and
what
doesn't,
but
yeah
I'll
I'll,
be
in
touch
with
you
separately
and
about
this
and
make
sure
that
it's
reasonable
and
and
that
you
you're
aware
of
what
needs
to
be
done.
Obviously,
it'll
be
great
to
have
your
input.
A
H
Just
it's
just
one
last
thing:
I
I
remember
posted
a
request
on
after
la
on
the
last
february
meeting
issue:
66.
C
H
Two
structures:
I
think
that
we've
discussed
about
it
so
that
two
molecules
were
based.
No,
the
last
latest
latest
match
all
the
way,
yeah
that
that
two
structures-
oh.
H
So
so
do
you
think
we
can
just
send
it
for
biological
testing
and
crystallography
just
for
fun?
H
A
Think
biological
evaluation
is
less
of
a
a
challenge.
If
warwick
has
the
a
decent
throughput
assay
from
running,
but
we
have
to
have
a
good
justification
for
it,
which
I
think
you've
got
there,
but
yeah.
H
A
And
make
sure
that
they
they'd
be
happy
to
screen
them
if,
if
we
sent
them.
A
Okay,
all
good
thanks
very
much
everybody
for
coming
along
I'll,
try
and
post
some
updates
and
we'll
talk
to
each
other
offline
and
then
next
month,
okay,
nice,
to
see
everyone
all
right.
It's
good
bye!
Thank.