To date, publi
shed pharmacophore elucidation approache
s typically u
se a handful of data
set
s for validation: here, we have a
ssembled a data
set for 81 target
s, containing 960 ligand
s aligned u
sing their cocry
stallized protein target
s, to provide the experimental 鈥済old
standard鈥? The two-dimen
sional
structure
s are al
so a
ssembled to remove conformational bia
s; an ideal method would be able to take the
se
structure
s a
s input, find the common feature
s, and reproduce the bioactive conformation
s and their alignment
s to corre
spond with the X-ray-determined gold
standard alignment
s. Here we pre
sent thi
s data
set and de
scribe three objective mea
sure
s to evaluate performance: the ability to identify the bioactive conformation, the ability to identify and correctly align thi
s conformation for 50% of the molecule
s in each data
set, and the pharmacophoric field
similarity. We have applied thi
s validation methodology to our pharmacophore elucidation method FLAPpharm, that i
s publi
shed in the fir
st paper of thi
s serie
s and di
scu
ss the limitation
s of the data
set and objective
succe
ss criteria. Starting from two-dimen
sional
structure
s and producing unbia
sed model
s, FLAPpharm wa
s able to identify the bioactive conformation
s for 67% of the ligand
s and al
so to produce
succe
ssful model
s according to the
second metric for 67% of the Pharmbench data
set
s. In
spection of the un
succe
ssful model
s highlighted the limitation of thi
s root mean
square (rm
s)-derived metric,
since many were found to be pharmacophorically rea
sonable, increa
sing the overall
succe
ss rate to 83%. The PharmBench data
set i
s available at
scovery.com/PharmBench" class="extLink">http://www.moldiscovery.com/PharmBench, along with a web service to enable users to score model alignments coming from external methods in the same way that we have presented here and, therefore, establishes a pharmacophore elucidation benchmark data set available to be used by the community.