The best classification accuracy obtained by the LRSMs on all datasets ranks high compared to that obtained by the majority of music genre classification techniques, as listed in last rows of Table 1.
The 4th ranked prediction is the known GATA E-box motif complex [ 21], which was also identified in the above-described analysis of Duke DNase-seq datasets (ranked 7th in FiGATA E-box
For AQP4, we determined the overlap of the 100 highest correlated and 100 highest anti-correlated genes across all microarray datasets ranked by Pearson correlation (Additional file 3, Table S3).
The rule 'IF NFIL3 is highly expressed THEN CG' (with NFIL3 coding for the nuclear factor interleukin-3-regulated protein) was generated with high relevance from both the 'Jena' and the 'Berlin' datasets (ranked in third and fourth position, respectively; Table 5).
The rankings provided by the individual datasets (GIM rank, SGA rank and E-MAP rank) were used as reference values for the integrative detection using four different rank aggregation functions (minimum, maximum, product and Borda count).
The size of this dataset ranks second in the literature, only smaller than the private dataset of Facebook.
The combined dataset ranked genes relatively more consistently.
When a measure differed significantly from normality, we performed a within-dataset rank-normalization to attain normality and improve the suitability for principal components analysis (see Appendix S4 for details).
We thus modified the parallel path enumeration algorithm to only store the top m paths from any source to any TF in our dataset, ranking the paths by path weight.
To assess if the contribution of the eigensystems 2 and 3 to the gene expression profiles is associated with biological functions, we analyzed the Gene Ontology "biological process" annotations of all genes in either dataset, ranked on the loadings of the respective eigensystems 2-3.
99% error quantile (A99) Datasets RMS Rank Avgerr.
