Context-dependent transcriptional regulations between signal transduction pathways
© Hwang et al; licensee BioMed Central Ltd. 2011
Received: 12 August 2010
Accepted: 13 January 2011
Published: 13 January 2011
Cells coordinate their metabolism, proliferation, and cellular communication according to environmental cues through signal transduction. Because signal transduction has a primary role in cellular processes, many experimental techniques and approaches have emerged to discover the molecular components and dynamics that are dependent on cellular contexts. However, omics approaches based on genome-wide expression analysis data comparing one differing condition (e.g. complex disease patients and normal subjects) did not investigate the dynamics and inter-pathway cross-communication that are dependent on cellular contexts. Therefore, we introduce a new computational omics approach for discovering signal transduction pathways regulated by transcription and transcriptional regulations between pathways in signaling networks that are dependent on cellular contexts, especially focusing on a transcription-mediated mechanism of inter-pathway cross-communication.
Applied to dendritic cells treated with lipopolysaccharide, our analysis well depicted how dendritic cells respond to the treatment through transcriptional regulations between signal transduction pathways in dendritic cell maturation and T cell activation.
Our new approach helps to understand the underlying biological phenomenon of expression data (e.g. complex diseases such as cancer) by providing a graphical network which shows transcriptional regulations between signal transduction pathways. The software programs are available upon request.
Signal transduction is the primary process by which cells coordinate their metabolism, proliferation, and cellular communication according to environmental signals such as hormones, nutrients, and other chemical stimuli. Cells sense environmental signals by receptor proteins which convert the signals into various responses through signal transduction that are dependent on cellular contexts such as signals, receptor proteins that cells possess, and intracellular machinery by which cells integrate and interpret the signals . For example, the JAK-STAT signal transduction pathway, which provides one of the most direct routes from cell-surface receptors to a nucleus, is activated by more than 30 cytokines of soluble mediators in cell communication. The cellular responses are different according to their cytokines even though they are stimulated by the same JAK-STAT signal transduction pathway .
As well as for various responses stimulated by signal transduction pathways or signaling pathways, recent articles have presented abundant evidence for inter-pathway cross-communication according to cellular contexts [2–4]. Cytokine signaling which is critical in immune system regulates functions of other signaling pathways either by transcription-mediated consequences of cytokine signaling or by transcription-independent mechanisms . As an example of transcription-mediated mechanisms, interferon gamma activates signal transduction pathways of toll-like receptors (TLRs) by inducing expression of TLRs . An example of transcription-independent mechanisms, Bezbradica and Medzhitove  suggested that lateral interactions between cytokine receptors and other cellular receptors may explain how different cells induce their cell-type specific responses with a highly limited set of janus kinase (JAK) and signal transducer and activator of transcription (STAT) signaling proteins.
Among the two mechanisms of cross-communication between signaling pathways according to cellular contexts, we focus on the transcription-mediated mechanism that can be inferred by integrating omics data as well as genome-wide expression data. Various methods analyzing expression data by integrating omics data have been employed to infer sub-networks perturbed at cellular context with protein-protein interaction (PPI) data[6–10]. Ideker et al.  first proposed to identify sub-networks by devising an adequate scoring function on PPI networks based on the significant changes in gene expression. By adapting the scoring concept, many similar approaches have improved the search algorithms  or scoring functions [6, 9]. However, previous approaches that inferred sub-networks did not provide transcription-mediated communication between signaling pathways, because they could not identify signaling pathways regulated by transcription at cellular contexts and PPI data have the noise problem .
Therefore, we propose a new computational omics approach for discovering signaling pathways regulated by transcription, Transcription-Regulating Signaling Pathways (TRS Pathways) and transcriptional regulations between pathways in Transcription-Regulating Signaling Networks (TRS Networks) that are dependent on cellular contexts. In this approach, cellular contexts are restricted to the experimental condition of expression data. TRS Pathways are signaling pathways of which some proteins are regulated by transcription according to the context of expression data. Signaling pathways are chains of proteins relaying a signal from ligands or transmembrane proteins to transcription factors, or some proteins whose roles are clearly known such as caspase3 [1, 12]. Proteins regulated by transcription represent the mRNA expression levels of proteins at the context which are significantly changed. TRS Networks are sub-networks which result from transcriptional relation between TRS Pathways. Applied to dendritic cells treated with lipopolysaccharide, we found several biological facts and transcriptional regulations as examples of inter-pathway cross-communication, related to dendritic cell maturation and T cell activation.
Our system comprises three major steps: (1) constructing a human omics network from PPIs and Protein-DNA (PD) interactions; (2) identifying TRS Pathways by two strategies: three constraints to reduce the search space for TRS Pathways and designing a scoring function for TRS Pathways; (3) identifying the TRS Networks by a search algorithm. After explaining these three major steps of the system, we briefly explain the scoring function for TRS Networks, designed to compare with sub-networks inferred by previous methods, expression data analyses and kinase reaction annotations.
Constructing a human omics network
The two strategies for identifying TRS Pathways
Because finding the highest-scoring connected sub-networks like TRS Pathways in the entire interaction network is a NP-hard problem , we adopted two strategies. The first strategy is to reduce the search space for TRS Pathways by three constraints. The second is to find the pathways only with top scores by defining the scoring function for a TRS Pathway.
The first strategy for identifying TRS Pathways: three constraints to reduce the search space for TRS Pathways
The first strategy for identifying TRS Pathways is to reduce the search space of the entire network for TRS Pathways by three constraints. The three constraints are based on three previous approaches mining candidate signaling pathways from PPI data, given a pair of starting and ending proteins [12, 22, 23]. Firstly, we search for TRS Pathways relaying a signal from 1,728 start proteins to 479 end proteins. According to the definition of signaling pathways by which cells convert extracellular signals into cellular responses , start proteins are defined as ligand or transmembrane proteins; end proteins as transcription factor proteins or some proteins whose roles are clearly known in cells. 1,728 Start proteins and 479 end proteins were found by searching three databases: 1,310 trans-membrane proteins as start proteins from the Locate database ; 236 transcription factors as end proteins from the PD interaction data; 418 start (e.g. ligand proteins) and 243 end proteins (roles are clearly known in cells, e.g. caspase8) from the KEGG database . These 243 end proteins are terminal nodes of KEGG signaling pathways linked to other signaling pathways or biological processes such as apoptosis.
Secondly, because signals are transmitted from an extracellular region into a nucleus, we defined the right orders of pathways as the following: from an extracellular region to a plasma membrane, from a plasma membrane to cytoplasm and from cytoplasm to a nucleus. When we search for TRS Pathways, we remove edges whose directions are opposite to the right orders such as from a plasma membrane to an extracellular region. Removing the opposite edges can lead to the loss of some feedback regulation processes between PPIs, but is an indispensible step to make the complex PPI data into a simple signaling pathway model to deal with. Moreover, since we can identify the transcriptional feedback regulation processes from the PD interaction data, the processes can compensate the loss defect. The sub-cellular localization data were from Locate database . Lastly, we search for TRS Pathways whose path lengths are less than or equal to 10, since the path lengths of all signaling paths in the KEGG database  are distributed from 1 to 10 (Figure S2). We search the KEGG database for all signaling paths with the found start and end proteins, then count their path lengths.
The second strategy for identifying TRS Pathways: the scoring function and search algorithm
We sum the Z i over all |N| genes in a TRS Pathway to produce an aggregate Z score (Z N ) for a TRS Pathway. Then, in order to properly capture the connection between expression and network topology, we investigate whether the score Z N of a TRS Pathway in Eq. (3) is higher than expected relative to a random set of genes. We randomly take 100,000 samples from all gene sets of size k using a Monte Carlo approach and calculate their scores Z N . The mean μ k and standard deviation (SD) σ k for each k are estimated and the noise in the estimates is reduced using a sliding window average. Using these estimates, the corrected score S(N) is calculated. The corrected score of random TRS Pathways is guaranteed to have a mean of μ = 0 and SD σ = 1.
R(E) estimates how reliable PPIs in a TRS Pathway are. Estimating the PPI reliability R(E) of a TRS Pathway is absolutely necessary to indentify the signaling pathways consisting of PPIs , because the PPI data are noisy . In this study, we adopt Bebek and Yang's method based on a logistic regression model . The model represents the probability of a true interaction as a function of four observed random variables on a pair of proteins: (1) the observed number of papers in which the interaction between two proteins was observed, (2) the Pearson correlation coefficient of expression measurements of the corresponding genes, (3) the proteins' small world clustering coefficient , and (4) the binary (0/1) protein subcellular localization data of interacting proteins . When interacting proteins co-localize in the same subcellular location, we give 1 to the interacting protein pair.
To search for the top path score TRS Pathways, we used Dijkstra's algorithm . Although we reduce the search space for TRS Pathways by three constraints, it is still impossible to search for all possible TRS Pathways in a few days. Also, we do not have to find them all. However, other alternative pathways of the same start and end nodes may also be important, just as the highest-scoring path in cell biology. Therefore, we slightly modified Dijkstra's algorithm, which searches for only the smallest-scoring path, to identify TRS Pathways with several top path scores. The statistical significance of TRS Pathways is measured by randomly permutating the expression of individual genes and performing the same search for TRS Pathways 1,000 times.
Identification of TRS Networks by a search algorithm
The following is the pseudo code of a search algorithm for a TRS Network. Input: an entire omics network G, all TRS Pathways relevant to the context, and a start gene. If the start gene is not given, the start gene is determined as the start node of the highest scoring TRS Pathway.
Initialize Gp by the top ranked TRS Pathway whose start node is the start gene
Add the end node of the initial TRS Pathway into Queue;
WHILE end nodes exist in Queue
POLL an end node e,
FOR i = 1 to n of outgoing edges of e;
IF a target node ti of ei is a DEG (Differentially Expressed Gene) or exists in Gp, add ti and ei into Gp;
IF added ti is a start node, a DEG, and positively regulated, add the top ranked TRS Pathways of ti into Gp and add the end node of the top ranked TRS Pathways into Queue;
IF added ti is an end and DEG, add ti into Queue;
Output G p and its score.
The above search algorithm represents the following; by the top ranked TRS Pathway, the signal is transmitted into the end nodes. The activated end nodes carry out their own roles in cells; transcription feedback regulation, enzyme activity or inducing the expression of new start proteins. The newly activated start proteins can send signals to other end nodes, thus we added the TRS Pathways activated by start proteins of a positively regulated DEG into the TRS Network. Since there are so many TRS Pathways at context, we have to select only the highly significant TRS Pathways for adding a TRS Network. The newly added end nodes also carry out their own roles in cells. Therefore, we repeat this process until there is no protein in the Queue.
The scoring function for TRS Networks for sub-network comparison
In order to calculate S , we use the same function for the expression score of a TRS Pathway (See Eq. (3)). P i is the score value (See Eq. (1)) of the ith TRS Pathway in a TRS Network. N i and E i also represent nodes and edges of ith TRS Pathway. N0 and E0 represent a set of nodes and a set of edges which do not belong to any TRS Pathway, respectively. They have a role to inter-connect TRS Pathways. To assess the statistical significance of TRS Networks, we take the same procedure in estimating the TRS Pathways by randomly permutating the expression of individual genes and performing the same search for TRS Networks 1,000 times.
Expression data analysis
We investigated dendritic cells treated with lipopolysaccharide (LPS), one of the TLR agonists. TLRs are very important innate receptors that sense microbial products and trigger dendritic cell maturation and cytokine production, effectively bridging innate and adaptive immunity . We downloaded GSE2706 data from the GEO database  and analyzed the data with limma . The limma results were used in three different ways. Firstly, we used the probability values (p values) of genes for the scoring function. Secondly, we selected genes for which the probability value was p < 0.05 as differential expressed genes for the identification of TRS Networks by a search algorithm. When we compared the performance of the TRS Network identification according to several DEG cut-off p values, p < 0.05 was one of the best performances (Figure S4). Lastly, we used fold changes of gene expression for the color of protein nodes in the networks.
Kinase reaction annotation
Since phosphorylation is one of the most important and critical processes in the signal transduction pathway, we annotate TRS Networks with the relationships between kinases and their substrates [33, 34]. We believe that the kinase reaction annotation makes the TRS Networks more reliable and more meaningful for a biological understanding.
TRS Pathways found by searching KEGG pathways and the human omics network
To find which signaling paths are regulated by transcription in our known knowledge, we searched the KEGG pathways  for TRS Pathways in dendritic cells 2h after the LPS treatment. The top TRS Pathways can be ranked by the two scores: the expression score and the path score (Additional file 1: Table S1). The expression score is an aggregated expression score S(N) (Eq. (3)) and the path score is our own path scoring function (Eq. (1)). Among all in the expression score, the path from TNF to MAPK12 (p38) was the highest-scoring and among all in the path score, the path from TNF to NFKB1 (p50) was the highest-scoring.
From the viewpoint of understanding the underlying biological phenomena of cellular contexts, the path from TNF to NFKB1 found by our path scoring function is more helpful, though it is difficult to determine which is more significant than the other. Let us suppose that intermediate signaling proteins exist to some extent in a cell to quickly respond to an environmental signal. Then, when the mRNA expression level of the start node of a path increases, we expect that the signal starting from the start node would be transmitted to the end protein of the path. Likewise if the mRNA expression level of the end node of a path increases, we also expect that the biological process of the end node would occur. In the path from TNF to NFKB, both the mRNA expression levels of the start node TNF and the end node NFKB1 were increased. The signal starting from TNF would be transmitted to NFKB1 and NFKB1 would activate the mRNA expression levels of its target genes as a transcription factor. Therefore, we can identify more explicable paths by using our path scoring function, adding the expression scores of the start and end node of a path (Z start and Z end of Eq. (2)) into the aggregated expression score, than by using just the aggregated score.
A TRS Pathway from TNF to NFKB1 was the highest-scoring when we searched the omics network with our path scoring function (See Eq. (1) and Methods). We compared it with the highest-scoring TRS Pathway of KEGG, since both pathways have the same start and end nodes (Figure 1). However, their intermediate nodes were shown to be quite different. Unless we thought of the actual biological system where a protein complex is usually used rather than a single protein for sending a signal, we could not recognize that the TRS Pathway of the omics network is also correct. The TRS Pathway of the omics network (Figure 1C) seems to be constructed by selecting each protein from its corresponding protein complex of Figure 1A to make a linear path.
The signaling pathway of Figure 1A has not been included in KEGG pathways, but it was already known in a paper . This shows that our TRS Pathway method suggests highly reliable TRS Pathway candidates by searching the omics network. Moreover, we also confirmed that our analysis can search not only highly reliable but also significantly regulated TRS Pathways from the omics network by assessing their significances quantitatively (See Methods). The TRS Pathway score (13.635) of the omics network (Figure 1C) was higher than that (11.936) of KEGG (Figure 1B), and the TRS Pathway p value (2.428e-4) of the omics network was lower than that (1.476e-3) of KEGG (Additional file 1: Table S2).
The first pathway sending a signal from LPS to the nuclear factor of kappa light polypeptide gene enhancer in B-cells (NFKB) was unfortunately hardly regulated by transcription. Not only was the path score low, but so was its expression score that we could not identify this pathway as a TRS Pathway. It was the correct result because the first pathway was not a TRS Pathway but a relevant signaling pathway in this context. These results shows that our TRS Pathway analysis can search highly reliable TRS Pathway candidates regulated by transcription according to cellular contexts well, even though it cannot guarantee to find all the relevant signaling pathways to the contexts. In our analysis, we fortunately knew that TLR4 senses LPS as the start protein that initiates all relevant signaling pathways. Therefore, we could identify transcriptional regulations between the first pathway and other signaling pathways in TRS Networks, though not when searching for TRS Pathways.
TRS Networks found by searching the omics network
We found that the TLR signaling pathway and Apoptosis are the most relevant and known pathways among all the KEGG pathways 2h after the LPS treatment, by Impact analysis . Impact analysis is one of the best methods ordering known pathways by estimating how each pathway is over-represented in a specific context. However, a major drawback to analyses using known pathways is that the large number of genes are uninvolved in pathways or have no functional classification . Even though we discovered the over-represented pathways in the context of expression data, we could not sometimes find out the new understanding of the biological phenomenon. Therefore, these two pathways are not the TRS Networks we wanted to discover but the most relevant and known pathways in this context.
The TLR4 TRS Network and TNF TRS Network were named by the start node used to search for TRS Networks (See Methods). To estimate the performance of our TRS Network analysis without knowing the start node, we searched for the two TRS Networks and compared their overlaps. In this context, TLR4 senses LPS as the start protein. If, however, we did not know the start protein, TNF, the start protein of the highest-scoring TRS Pathway would be selected as the start node of the search step. The TLR4 TRS Network was found by the actual start node and the TNF TRS Network was found by the inferred start node. The overall propensity of the overlaps of two networks looked very similar, though the performance by the actual start node was slightly better than by the inferred start node (Figure 2 and Additional file 1: Table S3). Therefore, this result shows that our TRS Network analysis works quite well on searching for TRS Networks without a known start protein.
The results of TRS Network analysis.
Networks or pathways
Network score (p value)
TLR4 TRS Network
TNF TRS Network
TLR signaling pathway
Among 70 KEGG pathways and the four sub-networks, the highest-scoring pathway or sub-network was the TLR4 TRS Network (Table 1). It includes more significant TRS Pathways (P : 942.988) than the TNF TRS Network (P : 698.589) and the TLR signaling pathway (P : 32.789), though the number of its edges (| |: 218) is similar to that of the TNF TRS Network (| |: 214) and to that of the TLR signaling pathway (| |: 213). This suggests that edges of the TLR4 TRS Network have a higher probability in being the components of significant TRS Pathways than those of other networks. Thus, the TLR4 TRS Network is more explicable and the edges may be more important. The D2D network and jActiveModule network did not obtain high network scores despite their high expression scores (S : 41.025 and 29.244). The networks comprise many edges (| |: 834 and 970) but do not include a lot of TRS Pathways (P : 19.484 and 32.089). This shows that most edges are not components of TRS Pathways. The two sub-networks are inexplicable and do not provide many clues to understand biological phenomenon of the context. Therefore, the orders of the network scores (Table 1) shows that our TRS Network analysis efficiently finds more TRS Pathways and TRS Networks than other approaches (including canonical pathway analyses).
TRS Network analysis shows how dendritic cells respond to the LPS treatment
Firstly, NFKB has gotten a positive feedback by inducing TNF, which in turn reinforces the activation of NFKB (Figure 3A). When the dendritic cells are treated with LPS, LPS activates TLR4 which recruits adaptor proteins and sends a signal to NFKB to release NFKB into the nucleus (a TLR4-NFKB pathway). NFKB in the nucleus induces the transcription of genes that promote immune and inflammatory responses . TNF, one of the target genes increased by NFKB, has the possibility to activate NFKB as a positive feedback by sending a signal to NFKB through a TNF-NFKB pathway. A positive feedback is possible because the TLR4-NFKB pathway and the TNF-NFKB pathway have a common downstream activating NFKB.
Secondly, NFKB activated by the TNF-NFKB pathway induces inflammatory responses by increasing the mRNA expression levels of inflammatory cytokines (e.g. IL6 and IL12B) (Figure 3B and Figure S5). The inflammatory cytokines IL6 and IL2B stimulates the JAK-STAT signaling pathway. The JAK-STAT signaling pathway provides one of the most direct routes to the nucleus in which transcriptional activation is initiated by each particular member of the STAT family . In spite of the simple pathway that consists of four JAKs and seven STATs, the pathway translates more than 30 cytokines into cell-type specific or context-dependent patterns of cytokine responsiveness and gene expression . In this context, cytokine IL6, IL12B, and OSM were activated by the TNF-NFKB pathway and the cytokines increased the mRNA expression levels of the interferon regulatory transcription factor family genes, such as IRF1 and IRF7 by the JAK-STAT signaling pathway. The chosen genes were transcription factors involved in inflammation and apoptosis . This inter-pathway cross-communication is a good example of transcriptional regulations between signaling pathways.
Thirdly, NFKB elevated the mRNA expression levels of necessary genes for T cell stimulation: CD40, CD80, ICAM1, CD83, CXCL10, CCL5, and CXCL11 (Figure S5). CD40 and CD80 are co-stimulatory molecules that bind to complementary receptors on the T cell surface, in activating a T cell . ICAM1 enables a T cell to remain bound to an antigen-presenting cell long enough for the T cell to become activated by binding lfa-1 on the T cell surface . Moreover, it was reported that the mature dendritic cells expressed CD83 and high levels of CD40, CD80, and CD86. CXCL10, CXCL11, and CCL5 are chemoattractants that guide the migrations of leukocytes such as T cells to induce a suitable immune response [43–45].
Lastly, NFKB increases the mRNA expression levels of genes related to apoptosis, especially inhibitors of apoptosis (IAPs) such as BIRC2 and BIRC3 suppressing caspases (Figure S5) . At 2h after the LPS treatment, mRNA expression levels of IAPs were only increased. However, at 8h after the treatment, those of other genes inhibiting apoptosis (IL15, NFKB1) were also increased. This result shows that apoptosis would be repressed in dendritic cells and it agrees with this context of the dendritic cell maturation. If apoptosis occurred in matured dendritic cells, they could not perform their roles initiating adaptive immune responses as an antigen-presenting cell.
We propose a new computational omics approach to discover signaling pathways regulated by transcription (TRS Pathways) and transcriptional regulations between them in TRS Networks dependent on cellular contexts to investigate the transcription-mediated mechanism of inter-pathway cross-communication of signaling pathways. Our new approach has three advantages. Firstly, highly reliable TRS Pathway candidates and transcriptional regulations between pathways can be discovered. Secondly, the approach can discover more known knowledge than the previous approaches. Lastly, it helps to understand the underlying biological phenomena of cellular contexts by providing a graphical network. We demonstrated that our analysis performed well in the context of dendritic cells treated with LPS, since dendritic cells commanding the human immune system are very important to disease research. Likewise, our approach determines how some signaling pathways are transcriptionally regulated by other pathways in patients by analyzing expression data comprising patients and normal samples. Therefore, it would be very helpful to understand the underlying biological phenomena of complex diseases such as cancer.
We thank Eunjung Lee, Taekwon Son and Sang Cheol Kim for biological and statistical consultation. We also thank Moonhee Lee for improving the manuscript. This work was supported by the Korean Systems Biology Research Project (20100002164), the World Class University program (R32-2008-000-10218-0) and the Basic Research Laboratory grant (2009-0086964) of the Ministry of Education, Science and Technology through the National Research Foundation of Korea.
- Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P: Molecular Biology of the Cell. 4th edition. 2002.Google Scholar
- Bezbradica JS, Medzhitov R: Integration of cytokine and heterologous receptor signaling pathways. Nat Immunol 2009, 10: 333–339. 10.1038/ni.1713View ArticlePubMedGoogle Scholar
- Ivashkiv LB: Cross-regulation of signaling by ITAM-associated receptors. Nat Immunol 2009, 10: 340–347. 10.1038/ni.1706PubMed CentralView ArticlePubMedGoogle Scholar
- Wilson NS, Dixit V, Ashkenazi A: Death receptor signal transducers: nodes of coordination in immune signaling networks. Nat Immunol 2009, 10: 348–355. 10.1038/ni.1714View ArticlePubMedGoogle Scholar
- Schroder K, Sweet MJ, Hume DA: Signal integration between IFNgamma and TLR signalling pathways in macrophages. Immunobiology 2006, 211: 511–524. 10.1016/j.imbio.2006.05.007View ArticlePubMedGoogle Scholar
- Guo Z, Li Y, Gong X, Yao C, Ma W, Wang D, Li Y, Zhu J, Zhang M, Yang D, Wang J: Edge-based scoring and searching method for identifying condition-responsive protein-protein interaction sub-network. Bioinformatics 2007, 23: 2121–2128. 10.1093/bioinformatics/btm294View ArticlePubMedGoogle Scholar
- Hwang S, Son SW, Kim SC, Kim YJ, Jeong H, Lee D: A protein interaction network associated with asthma. J Theor Biol 2008, 252: 722–731. 10.1016/j.jtbi.2008.02.011View ArticlePubMedGoogle Scholar
- Ideker T, Ozier O, Schwikowski B, Siegel AF: Discovering regulatory and signalling circuits in molecular interaction networks. Bioinformatics 2002, 18(Suppl 1):S233–240.View ArticlePubMedGoogle Scholar
- Rajagopalan D, Agarwal P: Inferring pathways from gene lists using a literature-derived network of biological relationships. Bioinformatics 2005, 21: 788–793. 10.1093/bioinformatics/bti069View ArticlePubMedGoogle Scholar
- Sohler F, Hanisch D, Zimmer R: New methods for joint analysis of biological networks and expression data. Bioinformatics 2004, 20: 1517–1521. 10.1093/bioinformatics/bth112View ArticlePubMedGoogle Scholar
- Deane CM, Salwinski L, Xenarios I, Eisenberg D: Protein interactions: two methods for assessment of the reliability of high throughput observations. Mol Cell Proteomics 2002, 1: 349–356. 10.1074/mcp.M100037-MCP200View ArticlePubMedGoogle Scholar
- Bebek G, Yang J: PathFinder: mining signal transduction pathway segments from protein-protein interaction networks. BMC Bioinformatics 2007, 8: 335. 10.1186/1471-2105-8-335PubMed CentralView ArticlePubMedGoogle Scholar
- Stark C, Breitkreutz BJ, Reguly T, Boucher L, Breitkreutz A, Tyers M: BioGRID: a general repository for interaction datasets. Nucleic Acids Res 2006, 34: D535–539. 10.1093/nar/gkj109PubMed CentralView ArticlePubMedGoogle Scholar
- Kerrien S, Alam-Faruque Y, Aranda B, Bancarz I, Bridge A, Derow C, Dimmer E, Feuermann M, Friedrichsen A, Huntley R, Kohler C, Khadake J, Leroy C, Liban A, Lieftink C, Montecchi-Palazzi L, Orchard S, Risse J, Robbe K, Roechert B, Thorneycroft D, Zhang Y, Apweiler R, Hermjakob H: IntAct--open source resource for molecular interaction data. Nucleic Acids Res 2007, 35: D561–565. 10.1093/nar/gkl958PubMed CentralView ArticlePubMedGoogle Scholar
- Peri S, Navarro JD, Kristiansen TZ, Amanchy R, Surendranath V, Muthusamy B, Gandhi TK, Chandrika KN, Deshpande N, Suresh S, Rashmi BP, Shanker K, Padma N, Niranjan V, Harsha HC, Talreja N, Vrushabendra BM, Ramya MA, Yatish AJ, Joy M, Shivashankar HN, Kavitha MP, Menezes M, Choudhury DR, Ghosh N, Saravana R, Chandran S, Mohan S, Jonnalagadda CK, et al.: Human protein reference database as a discovery resource for proteomics. Nucleic Acids Res 2004, 32: D497–501. 10.1093/nar/gkh070PubMed CentralView ArticlePubMedGoogle Scholar
- Chatr-aryamontri A, Ceol A, Palazzi LM, Nardelli G, Schneider MV, Castagnoli L, Cesareni G: MINT: the Molecular INTeraction database. Nucleic Acids Res 2007, 35: D572–574. 10.1093/nar/gkl950PubMed CentralView ArticlePubMedGoogle Scholar
- Messina DN, Glasscock J, Gish W, Lovett M: An ORFeome-based analysis of human transcription factor genes and the construction of a microarray to interrogate their expression. Genome Res 2004, 14: 2041–2047. 10.1101/gr.2584104PubMed CentralView ArticlePubMedGoogle Scholar
- Ryu T, Jung J, Lee S, Nam HJ, Hong SW, Yoo JW, Lee DK, Lee D: bZIPDB: a database of regulatory information for human bZIP transcription factors. BMC Genomics 2007, 8: 136. 10.1186/1471-2164-8-136PubMed CentralView ArticlePubMedGoogle Scholar
- Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, Pomeroy SL, Golub TR, Lander ES, Mesirov JP: Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci USA 2005, 102: 15545–15550. 10.1073/pnas.0506580102PubMed CentralView ArticlePubMedGoogle Scholar
- Kanehisa M, Araki M, Goto S, Hattori M, Hirakawa M, Itoh M, Katayama T, Kawashima S, Okuda S, Tokimatsu T, Yamanishi Y: KEGG for linking genomes to life and the environment. Nucleic Acids Res 2008, 36: D480–484. 10.1093/nar/gkm882PubMed CentralView ArticlePubMedGoogle Scholar
- Sayers EW, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S, Feolo M, Geer LY, Helmberg W, Kapustin Y, Landsman D, Lipman DJ, Madden TL, Maglott DR, Miller V, Mizrachi I, Ostell J, Pruitt KD, Schuler GD, Sequeira E, Sherry ST, Shumway M, Sirotkin K, Souvorov A, Starchenko G, et al.: Database resources of the National Center for Biotechnology Information. Nucleic acids research 2009, 37: D5–15. 10.1093/nar/gkn741PubMed CentralView ArticlePubMedGoogle Scholar
- Steffen M, Petti A, Aach J, D'Haeseleer P, Church G: Automated modelling of signal transduction networks. BMC Bioinformatics 2002, 3: 34. 10.1186/1471-2105-3-34PubMed CentralView ArticlePubMedGoogle Scholar
- Scott J, Ideker T, Karp RM, Sharan R: Efficient algorithms for detecting signaling pathways in protein interaction networks. J Comput Biol 2006, 13: 133–144. 10.1089/cmb.2006.13.133View ArticlePubMedGoogle Scholar
- Sprenger J, Lynn Fink J, Karunaratne S, Hanson K, Hamilton NA, Teasdale RD: LOCATE: a mammalian protein subcellular localization database. Nucleic Acids Res 2008, 36: D230–233. 10.1093/nar/gkm950PubMed CentralView ArticlePubMedGoogle Scholar
- Sharan R, Suthram S, Kelley RM, Kuhn T, McCuine S, Uetz P, Sittler T, Karp RM, Ideker T: Conserved patterns of protein interaction in multiple species. Proc Natl Acad Sci USA 2005, 102: 1974–1979. 10.1073/pnas.0409522102PubMed CentralView ArticlePubMedGoogle Scholar
- Goldberg DS, Roth FP: Assessing experimentally derived interactions in a small world. Proc Natl Acad Sci USA 2003, 100: 4372–4376. 10.1073/pnas.0735871100PubMed CentralView ArticlePubMedGoogle Scholar
- Huh WK, Falvo JV, Gerke LC, Carroll AS, Howson RW, Weissman JS, O'Shea EK: Global analysis of protein localization in budding yeast. Nature 2003, 425: 686–691. 10.1038/nature02026View ArticlePubMedGoogle Scholar
- Bader JS, Chaudhuri A, Rothberg JM, Chant J: Gaining confidence in high-throughput protein interaction networks. Nat Biotechnol 2004, 22: 78–85. 10.1038/nbt924View ArticlePubMedGoogle Scholar
- Cormen TH, Leiserson CE, Rivest RL, Stein C: Introduction to Algorithms. 2nd edition. MIT Press and McGraw-Hill; 2001.Google Scholar
- Iwasaki A, Medzhitov R: Toll-like receptor control of the adaptive immune responses. Nat Immunol 2004, 5: 987–995. 10.1038/ni1112View ArticlePubMedGoogle Scholar
- Barrett T, Troup DB, Wilhite SE, Ledoux P, Rudnev D, Evangelista C, Kim IF, Soboleva A, Tomashevsky M, Marshall KA, Phillippy KH, Sherman PM, Muertter RN, Edgar R: NCBI GEO: archive for high-throughput functional genomic data. Nucleic Acids Res 2009, 37: D885–890. 10.1093/nar/gkn764PubMed CentralView ArticlePubMedGoogle Scholar
- Smyth GK: Linear models and empirical bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004, 3: Article3.PubMedGoogle Scholar
- Linding R, Jensen LJ, Ostheimer GJ, van Vugt MA, Jorgensen C, Miron IM, Diella F, Colwill K, Taylor L, Elder K, Metalnikov P, Nguyen V, Pasculescu A, Jin J, Park JG, Samson LD, Woodgett JR, Russell RB, Bork P, Yaffe MB, Pawson T: Systematic discovery of in vivo phosphorylation networks. Cell 2007, 129: 1415–1426. 10.1016/j.cell.2007.05.052PubMed CentralView ArticlePubMedGoogle Scholar
- Manning G, Whyte DB, Martinez R, Hunter T, Sudarsanam S: The protein kinase complement of the human genome. In Science. Volume 298. New York, NY; 2002:1912–1934. 10.1126/science.1075762Google Scholar
- Bhoj VG, Chen ZJ: Ubiquitylation in innate and adaptive immunity. Nature 2009, 458: 430–437. 10.1038/nature07959View ArticlePubMedGoogle Scholar
- Draghici S, Khatri P, Tarca AL, Amin K, Done A, Voichita C, Georgescu C, Romero R: A systems biology approach for pathway level analysis. Genome Res 2007, 17: 1537–1545. 10.1101/gr.6202607PubMed CentralView ArticlePubMedGoogle Scholar
- Curtis RK, Oresic M, Vidal-Puig A: Pathways to the analysis of microarray data. Trends Biotechnol 2005, 23: 429–435. 10.1016/j.tibtech.2005.05.011View ArticlePubMedGoogle Scholar
- Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, Schwikowski B, Ideker T: Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 2003, 13: 2498–2504. 10.1101/gr.1239303PubMed CentralView ArticlePubMedGoogle Scholar
- Murray PJ: The JAK-STAT signaling pathway: input and output integration. J Immunol 2007, 178: 2623–2629.View ArticlePubMedGoogle Scholar
- El Bougrini J, Pampin M, Chelbi-Alix MK: Arsenic enhances the apoptosis induced by interferon gamma: key role of IRF-1. Cell Mol Biol (Noisy-le-grand) 2006, 52: 9–15.Google Scholar
- Dustin ML, Springer TA: T-cell receptor cross-linking transiently stimulates adhesiveness through LFA-1. Nature 1989, 341: 619–624. 10.1038/341619a0View ArticlePubMedGoogle Scholar
- Hayati AR, Zulkarnaen M: An immunohistochemical study of CD1a and CD83-positive infiltrating dendritic cell density in cervical neoplasia. Int J Gynecol Pathol 2007, 26: 83–88. 10.1097/01.pgp.0000225850.90115.bcView ArticlePubMedGoogle Scholar
- Cocchi F, DeVico AL, Garzino-Demo A, Arya SK, Gallo RC, Lusso P: Identification of RANTES, MIP-1 alpha, and MIP-1 beta as the major HIV-suppressive factors produced by CD8+ T cells. In Science. Volume 270. New York, NY; 1995:1811–1815. 10.1126/science.270.5243.1811Google Scholar
- Cole KE, Strick CA, Paradis TJ, Ogborne KT, Loetscher M, Gladue RP, Lin W, Boyd JG, Moser B, Wood DE, Sahagan BG, Neote K: Interferon-inducible T cell alpha chemoattractant (I-TAC): a novel non-ELR CXC chemokine with potent activity on activated T cells through selective high affinity binding to CXCR3. J Exp Med 1998, 187: 2009–2021. 10.1084/jem.187.12.2009PubMed CentralView ArticlePubMedGoogle Scholar
- Jinquan T, Jing C, Jacobi HH, Reimert CM, Millner A, Quan S, Hansen JB, Dissing S, Malling HJ, Skov PS, Poulsen LK: CXCR3 expression and activation of eosinophils: role of IFN-gamma-inducible protein-10 and monokine induced by IFN-gamma. J Immunol 2000, 165: 1548–1556.View ArticlePubMedGoogle Scholar
- Fisher RA: On the interpretation of χ2from contingency tables, and the calculation of P. J Royal Stat Soc 1922, 85: 87–94. 10.2307/2340521View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.