Skip to main content

Table 5 Analysis of the highly enriched terms (p-value below 0.01 after correction using Enrichr [33]) for the 182 pattern-based biclusters found with BicPAMS in the dlbcl dataset (human cellular responses to chemotherapy) against multiple repositories: pathway databases (KEEG, WIKI, Reactome and BioCarta), human PPIs, GO, NCI-60 and cancer cell line Encyclopedia, Human Gene Atlas and MSigDB

From: BicPAMS: software for biological data analysis with pattern-based biclustering

  Database Avg. terms (p <0.01) per bicluster Summary
Pathways KEEG Pathways 23±11 Each of the 182 biclusters has a compact set of coherent and significantly enriched pathways in the KEEG database. There is a high dissimilarity (low overlapping) of enriched pathways between biclusters. To illustrate the relevance of enriched pathways to characterize the putative biological role per bicluster, consider the following four discovered biclusters {BK1,BK2,BK3,BK4} with terms showing a corrected p-value below 1E-8. BK1 has enriched responses to antigens, including the signaling of FCER1 (controls the production of immune mediators) and NF-kappa pathways. BK2 shows enrichment of more global pathways associated with cancer and immunodeficiency. BK3 has enriched antigen processing and presentation, as well as pathways related with a diversity of autoimmune infections. BK4 is associated with B-cell receptor signaling as expected in chemotherapeutic regulation and pathways regulating the proliferation of (cancerous) cells.
  WIKI Pathways 20±7 Although dissimilarity of WIKI pathways between biclusters is also observed, the overlapping degree of pathways is higher than previous KEEG-based analysis. Consider the highly enriched terms (corrected p-value below 1E-8) for three randomly selected biclusters {BW1,BW2,BW3}. BW1 shows enriched signaling pathways associated B-Cell receptor, including signaling of type II interferon, TCR, almost all IL families, chemokine, and TSLP. BW2 has genes closely matching the genes associated with the B-Cell receptor signaling pathway. Finally, BW3 has enriched pathways involved in preventing cell proliferation (as expected after chemotherapy), including G1 to S cell cycle control.
  Reactome Pathways 69±37 The found biclusters have in average a higher number of enriched pathways in the Reactome than in peer databases. Considering two randomly selected biclusters {BR1,BR2} and pathways with enriched p-values below 1E-14 after correction. BR1 has enriched pathways associated with immune responses and B-signalings, including cytokine signaling in immune system, interferon signaling, adaptive immune system and immunoregulatory interactions between lymphoid and non-lymphoid cells. BR2 has enriched pathways associated with antigen activation of B-cell receptor and control of cell proliferation (including mitotic G1-G1/S phases, and G1/S and M/G1 transitions).
  BioCarta Pathways 5,5±2,5 The found biclusters are associated with small and dissimilar sets of enriched pathways in the BioCarta database. BioCarta provides unique pathway knowledge, being essential to guarantee a more complete view of the putative roles of biclusters. Let us consider the enriched pathways for 3 randomly selected biclusters, {BW1,BW2,BW3}. BW1 is associated with T-cell receptor (TCR) pathways, including TCR activation by tyrosine kinases, TCR apoptosis and TCR signaling. Similarly to WIKI pathways, BW2 is associated with the signaling of B-cell receptor (BCR) and BW3 with the control of cancerous cell proliferation (inc. regulation of DNA replication and p53 signaling).
Cell lines NCI-60 Cancer cell lines 5,3±2,1 The majority of biclusters shows a compact set of enriched cell lines – group of genes with unexpectedly high or low expression against remaining cell lines – with few overlapping cell lines between pairs of biclusters. This analysis is key t unravel unique properties of the lymphoma targeted by each bicluster. To illustrate, consider three randomly selected biclusters, {BN1,BN2,BN3}: BN1 was found to be primarily related with follicular lymphoma (RS11846 cell line with corrected 7.9E-9 p-value); BN2 was found to be associated with immunoblastic lymphoma (SR cell line with corrected 4.2E-10 p-value); and the {MOLT4,SW620,RPMI} cell lines enriched in BN3 (with corrected p-values below 1E-8) are associated with T-acute lymphoblastic leukemia, adenocarcinomas and chronic myelogenous leukemia.
  Cancer cell line Encyclopedia 47±30 The majority of enriched cancer cell lines were found to be associated with tumors of the hematopoietic and lymphoid tissues. In general, each bicluster shows an unique set of enriched cell lines. Consider 3 randomly selected biclusters {BC1,BC2,BC3} with enriched cell lines (corrected p-value below 1E-10): {DOHH2,KARPAS422,HS611T,WSUDLCL2,HT,SUDHL6} cell lines directly related with diffuse large B-cell lymphoma were enriched in BC1; {MOTN1,ALLSIL,MOLT16} cell lines related with (childhood) T acute lymphoblastic leukemia were enriched in BC2; and {HUT102,EHEB,JVM2} cell lines either pertaining to B-lymphoblastoid or mantle cell lymphoma were enriched in BC3.
Human Gene Atlas 4±1,4 The analysis of terms enriched in the human gene atlas is pertinent to understand the types of cells more likely to be affected by the putative biological responses modeled per bicluster. A few biclusters were found to be associated with effects on the whole blood cells, while the remaining majority of biclusters model more specific biological responses thus showing enrichment on specific types of cells. Considering four randomly selected biclusters {BH1,BH2,BH3,BH4}, we found 721 B lymphoblasts and CD19+ B cells (with p-values below 1E-6) associated with BH1, lymphoma burkitts (both Daudi and Raji with p-values below 7.2E-4) associated with BH2, CD14+ Monocytes, CD4+ Tcells, CD8+ Tcells (with p-values below 1E-4) associated with BH3, and CD33+ Myeloid and D56+ NKCells (with p-values below 1E-6) associated with BH4.
MSigDB Oncogenic Signatures 9±1,5 The Molecular Signatures database (MSigDB) tests the enrichment of genes with potential to cause cancer. Interestingly, the majority of the discovered biclusters have a single delineated oncogene (signature with considerably higher enrichment than peer signatures). A few illustrative signatures include: VEGFA UP with V1 DN (8.2E-8) corresponding to genes down-regulated by treatment with angionic factor VEGFA; RPS14 DN with V1 UP (4.3E-11) corresponding to genes up-regulated in CD34+ hematopoietic progenitor cells after knockdown of RPS14; or CAMP UP with V1 UP (3.4E-9) associated with genes up-regulated in primary thyrocyte cultures in response to cAMP signaling. This knowledge further discriminates the putative role of each bicluster.
Regulation Transcription factors 11±3 Compact and dissimilar sets of TFs were found to be associated with the found biclusters. Illustrating {STAT5A,STAT3,NFKB1}, {AIRE,ESR1,FOXP3,POU5F1,TP53} and {ILF2,CDKN1B,CCND1,UPF1} sets of TFs (with corrected p-values below 1E-3) were observed for three randomly selected biclusters. The analysis of TFs is essential to understand the putative regulatory mechanisms modeled by each bicluster. A more detailed analysis of enriched TFs per bicluster is provided in Additional file 1.
  PPI Hub Proteins 83±14 This analysis shows the proteins enriched per bicluster acting as hubs in interaction networks. Despite the large number of enriched hubs per bicluster, it is interesting to notice that biclusters show a low number of overlapping hub-proteins with each other. The analysis of four randomly selected biclusters revealed the {PTPN6,JAK2,CBL}, {GABARAPL1,GABARAPL2, GABARAP}, {SHC1,IL7R,SRC} and {MCC,SLC2A4,CDK1} sets of hub proteins with corrected p-values below 1E-10.
Gene Ontology GO Biological processes 298±90 All biclusters show a high number of functionally coherent terms associated with cellular biological processes. An analysis of the enrichment for some biclusters is provided in Table 6. Complementary analyzes are provided in Table 9.
  GO Cellular component 28±16 The analysis of the enriched cellular components provides complementary information to characterize the putative biological role of each bicluster. Given two randomly selected biclusters from the set of 182 biclusters: one bicluster was associated with cytosol and chromatin-related components (corrected p-values below 1E-10), while the other with cell surface and membranes (<1E-10). Unlike biological processes, a few pairs of biclusters share some cellular components.
  GO Molecular function 21±7 Similarly to cellular components, the knowledge of the enriched molecular functions can be used to enlarge the GO-based analysis of biclusters. Each bicluster was found to be associated with a compact set of molecular functions consistently related with the molecular mechanisms underlying immunological responses to chemotherapy. Considering two randomly selected biclusters: the first bicluster showed enriched terms (with corrected p-values below 1E-6) associated with antigen binding and the binding of amide, protein complexes and small proteins (inc. chemokine receptor); while the second bicluster showed enriched terms (<1E-6) associated with protein kinase binding and regulation, structure-specific DNA binding, and ATPase activity.