Figure 4.

Molecular fingerprinting sub-classify infratentorial from supratentorial PAs as well as separate mixed glial-neuronal tumours from PAs. a) heatmap - infratentorial vs. supratentorial PAs. Heatmap plot of the gene expression submatrix of the 27 PAs restricted to the 136 probesets selected by the l1l2 feature selection. The tumours are grouped according to the lesion site (infratentorial vs. supratentorial). Each column represents a sample and each row is associated to a probe-set. The relative expression of the probe-sets is normalized ranging from 0 (blue, under-represented) to 1 (red, over-expressed), b) This figure illustrates a 3-dimensional visualization of the dataset restricted to the 136 selected probe-sets. The 3D representation is obtained by projecting the data submatrix onto its 3 principal components i.e. the components of maximum variance. Red circles represent the supratentorial and the blue circles the infratentorial PAs; c) This figure illustrates a 3D projection of the dataset restricted to the 103 selected probe-sets for the supratentorial tumours: mixed glial-neuronal tumours vs. PAs. The 3D representation is obtained by projecting the data submatrix onto its 3 principal components i.e. the components of maximum variance. Red circles represent the PAs and the blue circles the mixed glial-neuronal tumours; d) Heatmap plot of the gene expression submatrix of the 22 supratentorial tumours restricted to the 103 probe-sets selected by the l1l2 feature selection. The tumours are grouped according to the histotype (mixed glial-neuronal tumours vs. PAs) Each column represents a sample and each row is associated to a probe-set. The relative expression of the probe-sets is normalized ranging from 0 (blue, under-represented) to 1 (red, over-expressed).

Mascelli et al. BMC Cancer 2013 13:387   doi:10.1186/1471-2407-13-387
Download authors' original image