Three genes, TRIL, NPCL1 and C4orf7 were selected by three of four of the feature selection methods. TRIL, was recently identified as a novel component of the TLR4 complex and TLR3 complex. TRIL mRNA expression has been detected in the small intestine, as well as the central nervous system, lung, kidney and ovary. TRIL expression is upregulated in cell culture by lipopolysaccharide. The upregulation of TRIL expression could reflect altered host microbial interactions in macroscopically disease unaffected regions of the intestine in ileal CD patients. NPC1L1 is required for intestinal uptake of Paclitaxel 33069-62-4 cholesterol and plant sterols and is relatively abundant in the ileum. Upregulation of NPC1L1 expression in ileal CD patients may also contribute to enhanced atherogenesis in Crohn’s patients. Partial correlation network analysis revealed that FOLH1 has nonzero correlations with 12 of the other 16 genes in the signature. The biological basis for the nonzero partial correlations between the “hub” gene, FOLH is not immediately apparent. We have previously noted downregulation of C4orf7 as well as other genes associated with organized lymphoid structures and/or B-cell function in ileal CD patients compared to non-IBD control patients. This study indicates that downregulation of ileal C4orf7 expression is also observed in ileal CD patients, when compared with UC patients. Partial correlation network analysis revealed that FOLH1 has nonzero correlations with 12 of the other 16 genes in the signature. The biological basis for the nonzero partial correlations between the “hub” gene, FOLH is not immediately apparent. Thus far, we have not detected association of the gene features listed above with alterations in microbial composition, but we are likely underpowered to detect such associations with only 81 samples with paired microbiome and microarray data. We also noted that upregulation of FOLH1 was observed in ileal CD samples regardless of NOD2 genotype.In this study we report the results of binary classification – ileal CD vs. non-CD. Our attempts to apply multiclassification to the data set yielded poor accuracy particularly between the UC and control non-IBD phenotypes. This may be partly because the number of UC samples and control non-IBD samples were both smaller than the number of ileal CD samples. Of note, the errors in the binary classification of ileal CD vs. non-CD reflected misclassification of two UC samples as ileal CD. In the original test set we had an additional sample from a subject with a pre-operative diagnosis of UC. However the post-operative diagnosis was changed to Crohn’s colitis based on the pathological diagnosis of the resected specimen. Interestingly this discarded sample was classified as “ileal CD” based on the expression profile. While the ileal CD phenotype can be easily distinguished from ulcerative colitis based on imaging and endoscopic findings, it is more difficult to distinguish Crohn’s colitis from ulcerative colitis even after pathological diagnosis of the resected colon.