PRELIMINARY-IMAGE This is a preliminary API until 2018.Oct and may be improved based on user feedback. It is currently available in C++ and Python.

class OEFPVariogram : public OEFPHistogram

OEFPVariogram class is used to generate a variogram of similarity scores and related empirical measurements. An OEFPVariogram object is a OEFPHistogram with additional variogram data regarding user provided measurements.

The empirical variogram has the form :

\(\hat{\gamma}(h)=\frac{1}{2|N(h)|}\sum\limits_{(i,j)\in N(h)} |z_i-z_j|^2\)

Where \(N(h)\) is the set of pairs that have a distance close to \(h\) and \(z_i\) is the measurement provided for the \(i\) th molecule.

For all of our built-in similarity measures, except for Tversky, distance is calculated a \(1-similarity\). It is not possible to calculate an empirical variogram using Tversky similarity.

The measurement in the context of a variogram can be any user provided value per fingerprint e.g. bioassay data or calculated properties.

The following methods are publicly inherited from OEFPHistogram:

AddSample GetCounts NumBins
GetBinBoundaries GetDensity NumSamples
GetBinCenters GetMax Std
GetBinIdx GetMin  
GetBinWidth Mean  


OEFPVariogram(size_t numBins=200, double min=0, double max=1)

Constructors that need documenting.


size_t AddSample(const float sim, const float obs)

Adds a sample representing a pair of molecules into the variogram

Similarity between fingerprints i and j
Measured difference between the relevant fingerprints \(z_i-z_j\)


Since we are dealing with \(|z_i-z_j|^2\) this relationship is actually symmetrical.


OESystem::OEIterBase<double> *GetVariogram() const

Returns an iterator over the empirical variogram.