Rcdk generates formulae from off-center mass?

gistfile1.r

library(rcdk)

limits <- list(
   c("C", 0, 20),
   c("H", 0, 24),
   c("N", 0, 1),
   c("O", 0,4))

f <- generate.formula(55.05423, 0.001, elements=limits, charge=1, validation=FALSE)
# generates the correct formula:
print(f)

rm(f)
try(f <- generate.formula(55.05423, 0.0005, elements=limits, charge=1, validation=FALSE))
# error

f <- generate.formula(55.05427, 0.0005, elements=limits, charge=1, validation=FALSE) # fail
f <- generate.formula(55.05428, 0.0005, elements=limits, charge=1, validation=FALSE) # works

try(f <- generate.formula(55.054275, 0.0005, elements=limits, charge=1, validation=FALSE)) # fail
f <- generate.formula(55.054276, 0.0005, elements=limits, charge=1, validation=FALSE) # works

f <- generate.formula(55.05453, 0.0005, elements=limits, charge=1, validation=FALSE)
# generates the correct formula, even though the actual formula mass is 0.5423 and outside
# of the bounds 55.05428 to 55.05478 (if they are two-sided)
print(f)

rm(f)
f <- generate.formula(55.05473, 0.0005, elements=limits, charge=1, validation=FALSE)
# even this gives the correct formula!
print(f)

# all the way up to:
rm(f)
f <- generate.formula(55.05523, 0.0005, elements=limits, charge=1, validation=FALSE)
print(f)

rm(f)


# Generate formulae from a matrix of input masses and windows
gf.string <- function(mass, limit) tryCatch(
  generate.formula(mass,limit, elements=limits, charge=1, validation=FALSE)[[1]]@string, error=function(e) NA)
gf.mass <- function(mass, limit) tryCatch(
  generate.formula(mass,limit, elements=limits, charge=1, validation=FALSE)[[1]]@mass, error=function(e) NA)


masses <- 55.05423 +seq(-0.001, 0.001, 0.0001)
mlimits <- c(0.0001, 0.0002, 0.0005, 0.001)

#tabulated <-as.data.frame(do.call(rbind, lapply(masses, function(m) unlist(lapply(mlimits, function(l) gf.string(m,l))))))
tabulated <-as.data.frame(do.call(rbind, lapply(masses, function(m) unlist(lapply(mlimits, function(l) gf.mass(m,l))))))
rownames(tabulated) <- masses
colnames(tabulated) <- mlimits

print(tabulated)

# -> generate.formula calculates formulas ignoring the charge=1 parameter (but returns the mass of the charged formula)!
# The former behaviour was to generate formulas from the charged mass (taking the charge into account). 
# i.e. generate.formula(55.05423, window=0.0001, charge=1) would return C4H7 at 55.05423