Build your own PiMap

PiMaps Toolbox turns a molecular collection into your own PiMap: a two-dimensional map where position carries chemical meaning. Molecules are grouped by hierarchical clustering on their molecular fingerprints or descriptors, producing clusters at successively finer levels. Below the clusters, the map is divided into cells, the finest partition, and every molecule is assigned to a cell with its own map coordinates. The result is a nested structure: the map contains clusters, clusters contain cells, and cells contain the individual molecules:

  • Position means something. Proximity on the map reflects structural proximity in descriptor space, not an arbitrary embedding.
  • Grows with your data. New molecules project onto an existing PiMap without rebuilding it, so the map you built stays the map you use.
  • One map, every analysis. Similarity, diversity selection, enrichment, and binning all reference the same PiMap.

Access

Available in two forms:

KNIME logo

KNIME extension

The same engine exposed as visual nodes

Click to explore ⌄

PiMaps icon: a grid of navy and yellow cells representing clustered molecular data points

Standalone CLI

Native executable, for scripting and pipeline integration.

Click to explore ⌄

What each command does

PiMaps CLI
Black command-line window showing "C:\PiMaps>pimaps builder" in white, with a grey badge showing the letter mu (μ) in the bottom-right corner.

PiMaps Builder creates a PiMap from a set of molecules, clustering them into cells and assigning each molecule its coordinates, cell number, and its cluster membership at every level of the hierarchy.

Black command-line window showing "C:\PiMaps>pimaps projection" in white, with a grey badge showing the capital letter pi (Π) in the bottom-right corner.

PiMaps Projection places a new set of molecules onto an existing PiMap without rebuilding it. It takes the new molecules and an existing map as inputs, and returns the molecules labeled with their assigned cell, alongside the original map, unchanged.

Black command-line window showing "C:\PiMaps>pimaps similarity" in white, on a single line, with a grey badge showing the capital letter sigma (Σ) in the bottom-right corner.

PiMaps Similarity finds the most similar molecules using their positions on the PiMap, either within a single set (intra) or between two sets (inter).

Black command-line window showing "C:\PiMaps>pimaps pca" in white, with a grey badge showing the capital letter gamma (Γ) in the bottom-right corner.

PiMaps PCA runs PCA separately inside each cell rather than across the whole map, so it captures local variation a global PCA would miss.

Black command-line window showing the command "C:\PiMaps>pimaps diversity --mode round_robin" split across two lines, with a grey badge showing the capital letter delta (Δ) in the bottom-right corner.

PiMaps Diversity picks a diverse subset of molecules from within each cell, so every region of the map stays represented.

Black command-line window showing "C:\PiMaps>pimaps bins" in white, with a grey badge showing the lowercase letter beta (β) in the bottom-right corner.

PiMaps BINS groups molecules within a cell into numbered bins from a list of molecular pairs. Each pair is treated as a link between two molecules, so molecules connected directly or indirectly fall into the same bin.

Black command-line window showing "C:\PiMaps>pimaps enrichment" in white, with a grey badge showing the lowercase letter epsilon (ε) in the bottom-right corner.

PiMaps Enrichment finds molecules with too few similar neighbors (“orphans”) and pairs them with similar molecules drawn from a second collection in the same region of the map.

What each node does

PiMaps Builder PiMaps BINS PiMaps Splitter PiMaps Similarity PiMaps Projection PiMaps PCA PiMaps Enrichement PiMaps Diversity Within Cells
PiMaps Builder PiMaps BINS PiMaps Splitter PiMaps Similarity PiMaps Projection PiMaps PCA PiMaps Enrichement PiMaps Diversity
Yellow KNIME node icon labeled with a white mu (μ), one input port and two output ports, green status light.

PiMaps Builder creates a PiMap from a set of molecules, clustering them into cells and assigning each molecule its coordinates, cell number, and its cluster membership at every level of the hierarchy.

Yellow KNIME node icon labeled with a white capital pi (Π), two input ports and one output port, green status light.

PiMaps Projection places a new set of molecules onto an existing PiMap without rebuilding it. It takes the new molecules and an existing map as inputs, and returns the molecules labeled with their assigned cell, alongside the original map, unchanged.

Yellow KNIME node icon labeled with a white capital sigma (Σ), two input ports and one output port, green status light

PiMaps Similarity finds the most similar molecules using their positions on the PiMap, either within a single set (intra) or between two sets (inter).

Yellow KNIME node icon labeled with a white capital gamma (Γ), one input port and one output port, green status light.

PiMaps PCA runs PCA separately inside each cell rather than across the whole map, so it captures local variation a global PCA would miss.

PiMaps Diversity picks a diverse subset of molecules from within each cell, so every region of the map stays represented.

Yellow KNIME node icon labeled with a white epsilon (ε), two input ports and two output ports, green status light.

PiMaps Enrichment finds molecules with too few similar neighbors (“orphans”) and pairs them with similar molecules drawn from a second collection in the same region of the map.

Yellow KNIME node icon labeled with a white beta (β), one input port and two output ports, green status light.

PiMaps BINS groups molecules within a cell into numbered bins from a list of molecular pairs. Each pair is treated as a link between two molecules, so molecules connected directly or indirectly fall into the same bin.

Yellow KNIME node icon labeled with a white lowercase sigma (σ), two input ports and one output port, green status light.

PiMaps Splitter extracts one or several hierarchical partition levels from a PiMap, set explicitly, as a range, or generated automatically as powers of two, and returns the input molecules labeled with their cluster number at each requested level.