. 2021 Feb;39(2):215-224.

doi: 10.1038/s41587-020-0652-7. Epub 2020 Sep 14.

Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses

Joshua Broyde^#¹, David R Simpson^#², Diana Murray^#¹, Evan O Paull¹, Brennan W Chu¹, Somnath Tagore¹, Sunny J Jones¹, Aaron T Griffin¹, Federico M Giorgi³, Alexander Lachmann⁴, Peter Jackson^{5

6}, E Alejandro Sweet-Cordero⁷, Barry Honig^{8

9

10

11}, Andrea Califano^{12

13

14

15

16

17

18}

Affiliations

¹ Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA.
² Division of Pediatric Hematology/Oncology, Department of Pediatrics, UCSF Benioff Children's Hospital, San Francisco, CA, USA.
³ Cancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK.
⁴ Mount Sinai Center for Bioinformatics; Department of Pharmacological Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
⁵ Baxter Laboratory, Department of Microbiology & Immunology, Stanford University, Palo Alto, CA, USA.
⁶ Department of Pathology, Stanford University, Palo Alto, CA, USA.
⁷ Division of Pediatric Hematology/Oncology, Department of Pediatrics, UCSF Benioff Children's Hospital, San Francisco, CA, USA. alejandro.sweet-cordero@ucsf.edu.
⁸ Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA. bh6@cumc.columbia.edu.
⁹ Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹⁰ Department of Medicine, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹¹ Zuckerman Mind Brain and Behavior Institute, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹² Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹³ Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁴ Department of Medicine, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁵ JP Sulzberger Columbia Genome Center, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁶ Department of Biomedical Informatics, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁷ Institute for Cancer Genetics, Herbert Irving Comprehensive Cancer Center, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁸ Motor Neuron Center and Columbia Initiative in Stem Cells, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.

^# Contributed equally.

PMID: 32929263
PMCID: PMC7878435
DOI: 10.1038/s41587-020-0652-7

Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses

Joshua Broyde et al. Nat Biotechnol. 2021 Feb.

. 2021 Feb;39(2):215-224.

doi: 10.1038/s41587-020-0652-7. Epub 2020 Sep 14.

Authors

Affiliations

¹ Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA.
² Division of Pediatric Hematology/Oncology, Department of Pediatrics, UCSF Benioff Children's Hospital, San Francisco, CA, USA.
³ Cancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK.
⁴ Mount Sinai Center for Bioinformatics; Department of Pharmacological Sciences, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
⁵ Baxter Laboratory, Department of Microbiology & Immunology, Stanford University, Palo Alto, CA, USA.
⁶ Department of Pathology, Stanford University, Palo Alto, CA, USA.
⁷ Division of Pediatric Hematology/Oncology, Department of Pediatrics, UCSF Benioff Children's Hospital, San Francisco, CA, USA. alejandro.sweet-cordero@ucsf.edu.
⁸ Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA. bh6@cumc.columbia.edu.
⁹ Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹⁰ Department of Medicine, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹¹ Zuckerman Mind Brain and Behavior Institute, Columbia University, New York, NY, USA. bh6@cumc.columbia.edu.
¹² Department of Systems Biology, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹³ Department of Biochemistry and Molecular Biophysics, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁴ Department of Medicine, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁵ JP Sulzberger Columbia Genome Center, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁶ Department of Biomedical Informatics, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁷ Institute for Cancer Genetics, Herbert Irving Comprehensive Cancer Center, Columbia University Irving Medical Center, New York, NY, USA. ac2248@cumc.columbia.edu.
¹⁸ Motor Neuron Center and Columbia Initiative in Stem Cells, Columbia University, New York, NY, USA. ac2248@cumc.columbia.edu.

^# Contributed equally.

PMID: 32929263
PMCID: PMC7878435
DOI: 10.1038/s41587-020-0652-7

Abstract

Tumor-specific elucidation of physical and functional oncoprotein interactions could improve tumorigenic mechanism characterization and therapeutic response prediction. Current interaction models and pathways, however, lack context specificity and are not oncoprotein specific. We introduce SigMaps as context-specific networks, comprising modulators, effectors and cognate binding-partners of a specific oncoprotein. SigMaps are reconstructed de novo by integrating diverse evidence sources-including protein structure, gene expression and mutational profiles-via the OncoSig machine learning framework. We first generated a KRAS-specific SigMap for lung adenocarcinoma, which recapitulated published KRAS biology, identified novel synthetic lethal proteins that were experimentally validated in three-dimensional spheroid models and established uncharacterized crosstalk with RAB/RHO. To show that OncoSig is generalizable, we first inferred SigMaps for the ten most mutated human oncoproteins and then for the full repertoire of 715 proteins in the COSMIC Cancer Gene Census. Taken together, these SigMaps show that the cell's regulatory and signaling architecture is highly tissue specific.

PubMed Disclaimer

Conflict of interest statement

Conflicts of Interest

A.C. is founder and equity holder of DarwinHealth Inc., a company that has licensed some of the algorithms used in this manuscript from Columbia University. Columbia University is also an equity holder in DarwinHealth Inc.

Figures

**Extended Data Figure 1. The top 40 predictions from the OncoSig_NB algorithm for the KRAS LUAD SigMap were chosen for validation.**
**(a):** Performance of OncoSig_NB at recovering the Ingenuity-derived PGSS as a function of LR_Post. At an LR_Post = 53 (probability = 0.50) (vertical gray line), the OncoSig_NB LUAD-specific KRAS SigMap contains 10% of the PGSS (horizontal gray line). The vertical red line corresponds to LR_Post = 240, the cutoff used to obtain candidates for experimental validation. **(b):** ROC curve analysis, evaluated as the recovery of the Ingenuity-derived PGSS (FPR ≤ 0.05), for 1) OncoSig_NB (green curve, N = 1028)), 2) Pearson’s correlation between mRNA expression of KRAS and mRNA expression of other proteins in LUAD (blue curve), and 3) random performance (black curve). Recovery using 2-fold cross-validation (green) is essentially indistinguishable from recovery using 100-fold Monte-Carlo Cross-validation (not shown). 393 OncoSig_NB LUAD-specific KRAS SigMap predictions are made for LR_Post ≥ 53, which corresponds to probability ≥ 0.50 and FPR ≤ 0.019 (purple dot). 40 OncoSig_NB LUAD-specific KRAS SigMap predictions are made for LR_Post ≥ 240, which corresponds to probability ≥ 0.82 and FPR ≤ 0.0018 (yellow dot). The top 40 predictions are listed by gene name in **(c)**. **(c):** Orange and blue boxes contain, respectively, known upstream regulators and downstream effectors that are successfully recovered by OncoSig_NB. Italicized text indicates proteins known to interact with KRAS via a physical protein-protein interaction. The box titled “validated predictions” shows the novel OncoSig_NB predictions tested with the RNAi negative screen; those that were experimentally found to affect cell growth in a KRAS-dependent context are highlighted in bold text.

**Extended Data Figure 2. The OncoSig_RF and OncoSig_NB algorithms produce highly similar KRAS LUAD SigMaps.**
**(a)** Comparison of ROC curves (FPR ≤ 0.05) for LUAD-specific KRAS SigMaps predicted by OncoSig_NB (green and blue curves) and OncoSig_RF (orange and red curve) trained on the Ingenuity PGSS and the MSigDB PGSS, respectively. **(b)** Gene Set Enrichment Analysis (GSEA) of the top 100 OncoSig_NB LUAD-specific KRAS SigMap predictions at the top of the OncoSig_RF LUAD-specific KRAS SigMap predictions. Ranking is based on OncoSig_RF score (S_RF). Both the OncoSig_NB predictions tested in the knockdown experiments (red lines) and the remaining top 100 OncoSig_NB predictions (blue lines) are highly enriched at the top of the OncoSig_RF predictions (p = 5.6 x 10⁻⁸ and p = 1.7 x 10⁻¹⁹, respectively).

**Extended Data Figure 3. The log₂FC of shRNA abundance is plotted against the novel proteins tested in the KRAS negative selection screen.**
The 3-5 points plotted for a given protein are shRNAs that target the mRNA for that protein (N = 100). The X-axis is sorted by mean log₂FC for all shRNAs targeting each gene. Colors change from red to green with mean log₂FC.

**Extended Data Figure 4. OncoSig_RF predictions are highly enriched in oncogenic KRAS^Mut dependencies.**
**(a):** GSEA of KRAS^Mut synthetic lethal partners (25) (blue lines, N =216) and the top 500 OncoSig_RF LUAD-specific KRAS SigMap predictions obtained by training on a modified PGSS for which the intersection with the synthetic lethal set was removed. Inset is the GSEA using all OncoSig_RF predictions obtained in this way, where the ranking is OncoSig_RF score. Enrichment analysis was performed with the aREA (analytic Rank-based Enrichment Analysis) algorithm (10). **(b):** Enrichment of the protein resistance-signature to ERK inhibitor SCH772984 (28) (blue lines, N = 24) within OncoSig_RF LUAD-specific KRAS SigMap predictions. **(c):** Enrichment of proteins involved in response to Reactive Oxygen Species (GO:0000302) (8, 29) (blue lines, N = 276) within OncoSig_RF LUAD-specific KRAS SigMap predictions.

**Extended Data Figure 5. OncoSig_RF KRAS SigMaps exhibit tissue context specificity.**
**(a):** ROC curves for the OncoSig_RF KRAS SigMaps in LUAD (red), LUSC (gray) COAD (brown), and PAAD (orange) for FPR ≤ 0.05. Performance is evaluated as the recovery of established KRAS pathway proteins. **(b):** Gene set enrichment analysis (GSEA) of KRAS^Mut synthetic lethal partners, as determined by Corcoran et al. (27) (N = 48, blue lines). To avoid training and testing on the same proteins, OncoSig_RF predictions for COAD-specific KRAS SigMap proteins were obtained by training on a modified PGSS from which any established KRAS^Mut synthetic lethal protein had been previously removed. Enrichment analysis was performed with the aREA (analytic Rank-based Enrichment Analysis) algorithm. **(c):** Scatterplot of OncoSig_RF scores for KRAS SigMap proteins in PAAD-vs-LUAD (N = 19,789). Each dot represents the scores for one protein. Darker colored points have high scores (S_RF ≥ 0.5) in at least one context, and lighter colored points score poorly in both contexts (S_RF ≤ 0.5). R²_PAAD/LUAD = 0.037. **(d):** OncoSig_RF COAD-specific KRAS SigMap in the form depicted conceptually in Figure 1a. To prevent visual cluttering, only the top 33 OncoSig_RF predictions (FPR ≤ 0.01) that are also VIPER-inferred KRAS interactors (p ≤ 0.01), PrePPI-predicted KRAS physical interactors, or both, are depicted. Bold and regular text node labels represent established and novel predictions, respectively; orange and blue node colors represent upstream regulators and downstream effectors, respectively; red, blue, and black node borders represent predictions that are druggable (Drug Repurposing Hub (22)), KRAS^Mut synthetic lethal from the literature and validated here (see text), and both, respectively; orange and blue solid lines and gray nodes represent PrePPI-predicted physical interactors of KRAS.

**Extended Data Figure 6. OncoSig_RF SigMaps for hypermutated oncoproteins are retrospectively validated.**
**(a)** Pairwise overlap of established pathway proteins (left) and the OncoSig_RF LUAD-specific SigMaps (FPR ≤ 0.01, right) for the ten hyper-mutated oncoproteins (names of columns and rows). Percent overlap is color-coded according to the scale at top. **(b)** SigMap predictions are highly enriched in 600 EGFR-centric network proteins (52) (p = 2.3 x 10⁻⁴³). Enrichment analysis was performed with the aREA (analytic Rank-based Enrichment Analysis) algorithm. **(c)** Box plots of the OncoSig_RF LUAD-specific EGFR SigMap scores for two subsets of the curated EGFR pathway proteins from Astsaturov et al. (52): those identified as EGFR synthetic lethal partners (red, N = 58) and those not identified as synthetic lethal (grey, N = 542). The p-value (2 x 10⁻⁴) was calculated using Welch’s two sample t-test.

**Figure 1:. Protein-specific molecular interaction Signaling Map (SigMap) and the OncoSig_RF algorithm**
**(a)** Graphical representation of a SigMap for an anchor oncoprotein (red node). The SigMap comprises: (i) upstream activity modulators (orange nodes), (ii) downstream effectors, responsible for mediating its pathophysiologic function (blue nodes), (*iii*) structural cognate binding partners (gray nodes), which may be either modulators (solid orange lines) or effectors (solid blue lines), and (iv) auto-regulatory loops connecting downstream effectors to upstream modulators (dashed green lines). To avoid unnecessary clutter, implicit arrows connecting upstream modulators to the red node and the latter to its downstream effectors are omitted. Thus, the only interactions explicitly denoted by an edge are physical protein-protein interactions and autoregulatory interactions between modulator and effector proteins. **(b)** Networks used to train the OncoSig_RF algorithm. (i) PrePPI predicts interactions between a protein (red), and its physical and/or functional interactors (gray). (ii) The ARACNe algorithm predicts transcription factors or signaling molecules (red) that transcriptionally regulate target genes (blue). (*iii*) CINDy predicts signaling molecules (orange/red) that post-translationally modify transcription factors (blue boxes), which in turn leads to differential expression of a transcription factor’s targets (blue diamonds). (iv) The VIPER algorithm infers downstream effectors (blue) and upstream regulators (orange) for a given protein (red). VIPER associates 1) the protein (red) with a missense mutation (black dot) with the activity change of transcription factors (blue) and 2) signaling molecules (orange) with missense mutations (black dots) with activity of the protein (red). **(c)** Feature matrix for OncoSig_RF algorithm. Networks in (b) are encoded as a feature matrix (dark green box), where rows correspond to proteins in the human proteome, columns correspond to proteins for which clues exist in PrePPI, ARACNe, CINDy, and VIPER, respectively, and each entry is a scalar proportional to the confidence in the corresponding interaction as described in the literature. The latter include likelihood ratios for PrePPI, mutual information for ARACNe, number of SP-coF/TF-gene triplets for CINDy, and −log₁₀ p-value for VIPER. The gold column corresponds to whether a protein is an established member of a particular pathway (PGSS, value of 1) or not (value of 0). Only a few components of a small subset of proteins are shown. The feature vector for LIMD1 (highlighted in light green) is described in the text. The last column provides the OncoSig_RF score of the subset proteins for the LUAD-specific KRAS SigMap (see Figure 2). SP = signaling protein, coF = co-factor, and TF = transcription factor.

**Figure 2:. The OncoSig_RF LUAD-specific KRAS SigMap**
**(a):** ROC curves are displayed for the performance at recovering established KRAS-pathway proteins (FPR ≤ 0.05 (5%)) of OncoSig_RF (red curve, N = 1,114), Pearson’s correlation between mRNA expression of KRAS and mRNA expression of other proteins in LUAD (green curve, N = 957), and random prediction (black curve). The inset shows the full ROC curves (red curve, N = 19,548; green curve, N = 18,891). **(b):** The ROC curve in **(a)** for FPR ≤ 0.01 (1%, N = 263) is separated into two according to whether predictions correspond to **(i)** established KRAS-pathway proteins (top panel, yellow circles), with best-known KRAS-pathway proteins individually labeled or **(ii)** novel KRAS SigMap proteins (bottom panel, white circles). Circles annotate predictions as either druggable (Drug Repurposing Hub) (red), experimentally-validated KRAS interactors (BioGRID) (blue), or both (black). . **(c):** OncoSig_RF LUAD-specific KRAS SigMap in the form depicted conceptually in Figure 1a. To prevent visual cluttering, only the top 68 OncoSig_RF predictions that are also VIPER-inferred KRAS interactors, PrePPI-predicted physical interactors, or both, are depicted. Bold and regular text node labels represent established and novel predictions, respectively; orange and blue node colors represent upstream regulators and downstream effectors, respectively; red, black, and purple node borders represent predictions that are druggable (Drug Repurposing Hub), KRAS^Mut synthetic lethal partners from the literature or validated in this study, and both, respectively; solid orange and blue lines and gray nodes represent PrePPI-predicted physical KRAS interactors; green dashed lines represent auto-regulatory and feed-forward loop interactions.

**Figure 3:. Experimental validation of the OncoSig_RF LUAD-specific KRAS SigMap**
**(a):** Schematic of the pooled shRNA negative screen experiments performed. An average of four shRNAs target each gene in the protocol implemented. KRAS^G12D/+/p53^fl/fl primary tumor cells (green patches) are isolated from the mouse and placed in a semi-solid 3D matrix (cylinder). A pooled shRNA knockdown is performed (Day 1), and each cell stochastically integrates one shRNA into its DNA. Cells that integrate different shRNAs are shown as, red (representing shRNAs for novel predictions), green and purple (for positive controls), and black (for the background pool). Some cells and their daughter cells form spheroids (Day 6). The spheroids are dissociated, reseeded in a new matrix, and reform (Day 12). Fold Change (FC) of shRNA abundance is measured by deep sequencing the shRNAs at days 6 and 12. **(b):** Plot of log₂FC of shRNAs targeting predicted KRAS functional partners (red, N = 100), known members of the KRAS signaling pathways (RALGDS, MAPK1, RASA1 and AKT1) (purple, N = 17) and two synthetic lethal positive controls (NUP205 and TBK1) (green, N = 8). The black dots show log₂FC of shRNAs targeting 515 genes within the Background Pooled Screens (BPS, N = 2286) not expected to be involved in KRAS regulated signaling. The X-axis is the normalized rank, calculated by ranking log₂FC of each set of shRNAs and dividing by the number of shRNAs in that set. Each gene is represented by several dots, which correspond to different shRNAs. See Extended Data Figure 3 for more details. **(c)**: Density plots of log₂FC for predicted KRAS functional partners (red), all individual BPS (grey), and the average of all BPS (black). **(d):** Fold change versus significance for genes that significantly reduce organoid growth in the pooled shRNA negative screen experiments (FDR < 0.05, gray dotted line). Log₂FC of shRNAs, averaged after removing one or two outlier hairpins, is plotted against log₁₀-transformed, FDR-adjusted p-values. shRNAs are colored as described in **(b)**.

**Figure 4:. KRAS SigMap tumor context specificity**
Scatterplots of OncoSig_RF scores for KRAS SigMap proteins in **(a)** the LUSC-vs-LUAD (N = 19,790) and **(b)** the COAD-vs-LUAD (N = 19,438) tumor contexts **(b)**. Each dot represents one protein. Gold and gray points represent established and novel predictions, respectively. Darker colored points have high scores (S_RF ≥ 0.5) in at least one context, and lighter colored points have low scores (S_RF < 0.5) in both contexts and should thus not be compared. Correlation coefficients are R²_LUSC/LUAD = 0.35 (p < 10⁻²⁶⁷) and R²_COAD/LUAD = 0.10 (p < 10⁻¹⁶⁵), respectively (Welch’s Two Sample t-test). Specific points highlighted in black and green are discussed in the text.

**Figure 5:. OncoSig_RF LUAD-specific SigMap analysis of hyper-mutated oncoproteins**
**(a):** ROC curves showing OncoSig_RF’s performance in terms of identifying established pathway proteins in LUAD-specific SigMaps for the 10 oncoproteins listed in the legend (FPR ≤ 0.05, N ~ 1000). The thick red line represents performance of OncoSig_RF for the LUAD-specific KRAS SigMap from Figure 2a as a reference. **(b):** Number of literature-derived KRAS^Mut synthetic lethal partners (predicted by each of the ten SigMaps) as a function of OncoSig_RF score. OncoSig_RF scores are binned, and the bins are colored from dark red (highest scores) to white (S_RF < 0.50), as depicted in the leftmost column. The number of predictions per score bin is color-coded according to the legend at the top.

**Figure 6:. OncoSig: Generalization of OncoSig_RF to Cancer Gene Census proteins**
**(a):** ROC curve analysis of LUAD-specific KRAS SigMaps generated by OncoSig (blue curve) versus OncoSig_RF (red curve), FPR ≤ 0.10. Established KRAS signaling proteins recovered by OncoSig and OncoSig_RF are labeled in gray and black, respectively. **(b-d):** LUAD-specific OncoSig SigMaps for three Cancer Gene Census proteins in the form depicted conceptually in Figure 1a: SMARCA4 (b), MET (c), and BIRC6 (d). Orange and blue node colors represent upstream regulators and downstream effectors, respectively; solid orange and blue lines and gray nodes represent PrePPI-predicted physical interactors; red node borders represent predictions that are druggable (Drug Repurposing Hub); blue node text represents experimentally observed interactions (BioGRID); and dashed lines represent regulatory interactions. In **(b)**, the gray dashed box includes map members involved in chromatin organization (GO:0016568); green shading includes map members involved in histone acetylation (GO:0016573); and blue shading includes map members involved in chromatin remodeling (GO:0006338). In **(d)**, blue shading includes the map members involved in protein ubiquitination (GO:0016567), and green shading includes map members involved in protein sumoylation (GO:0016925).

See this image and copyright information in PMC

Cited by

Network-based elucidation of colon cancer drug resistance mechanisms by phosphoproteomic time-series analysis.
Rosenberger G, Li W, Turunen M, He J, Subramaniam PS, Pampou S, Griffin AT, Karan C, Kerwin P, Murray D, Honig B, Liu Y, Califano A. Rosenberger G, et al. Nat Commun. 2024 May 9;15(1):3909. doi: 10.1038/s41467-024-47957-3. Nat Commun. 2024. PMID: 38724493 Free PMC article.
Pancancer network analysis reveals key master regulators for cancer invasiveness.
Jethalia M, Jani SP, Ceccarelli M, Mall R. Jethalia M, et al. J Transl Med. 2023 Aug 20;21(1):558. doi: 10.1186/s12967-023-04435-6. J Transl Med. 2023. PMID: 37599366 Free PMC article.
UHRF1 is a mediator of KRAS driven oncogenesis in lung adenocarcinoma.
Kostyrko K, Román M, Lee AG, Simpson DR, Dinh PT, Leung SG, Marini KD, Kelly MR, Broyde J, Califano A, Jackson PK, Sweet-Cordero EA. Kostyrko K, et al. Nat Commun. 2023 Jul 5;14(1):3966. doi: 10.1038/s41467-023-39591-2. Nat Commun. 2023. PMID: 37407562 Free PMC article.
Identifying prognostic subgroups of luminal-A breast cancer using deep autoencoders and gene expressions.
Wang S, Lee D. Wang S, et al. PLoS Comput Biol. 2023 May 30;19(5):e1011197. doi: 10.1371/journal.pcbi.1011197. eCollection 2023 May. PLoS Comput Biol. 2023. PMID: 37253056 Free PMC article.
PrePPI: A Structure Informed Proteome-wide Database of Protein-Protein Interactions.
Petrey D, Zhao H, Trudeau SJ, Murray D, Honig B. Petrey D, et al. J Mol Biol. 2023 Jul 15;435(14):168052. doi: 10.1016/j.jmb.2023.168052. Epub 2023 Mar 17. J Mol Biol. 2023. PMID: 36933822 Free PMC article.

See all "Cited by" articles

References

1. Prahallad A et al. Unresponsiveness of colon cancer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature 483, 100–103, doi:10.1038/nature10868 (2012). - DOI - PubMed
1. Bild AH et al. Oncogenic pathway signatures in human cancers as a guide to targeted therapies. Nature 439, 353–357, doi:10.1038/nature04296 (2006). - DOI - PubMed
1. Krogan NJ, Lippman S, Agard DA, Ashworth A & Ideker T The cancer cell map initiative: defining the hallmark networks of cancer. Mol Cell 58, 690–698, doi:10.1016/j.molcel.2015.05.008 (2015). - DOI - PMC - PubMed
1. Greene CS et al. Understanding multicellular function and disease with human tissue-specific networks. Nat Genet 47, 569–576, doi:10.1038/ng.3259 (2015). - DOI - PMC - PubMed
1. Cancer Genome Atlas Research, N. et al. The Cancer Genome Atlas Pan-Cancer analysis project. Nat Genet 45, 1113–1120, doi:10.1038/ng.2764 (2013). - DOI - PMC - PubMed

Methods-only References

1. Woo JH et al. Elucidating Compound Mechanism of Action by Network Perturbation Analysis. Cell 162, 441–451, doi:10.1016/j.cell.2015.05.056 (2015). - DOI - PMC - PubMed
1. Duan Q et al. LINCS Canvas Browser: interactive web app to query, browse and interrogate LINCS L1000 gene expression signatures. Nucleic Acids Res 42, W449–460, doi:10.1093/nar/gku476 (2014). - DOI - PMC - PubMed
1. Kramer A, Green J, Pollard J Jr. & Tugendreich S Causal analysis approaches in Ingenuity Pathway Analysis. Bioinformatics 30, 523–530, doi:10.1093/bioinformatics/btt703 (2014). - DOI - PMC - PubMed
1. Li W & Godzik A Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics 22, 1658–1659, doi:10.1093/bioinformatics/btl158 (2006). - DOI - PubMed
1. Arlot S & Celisse A A survey of cross-validation procedures for model selection. Statist. Surv 4, 40–79, doi:doi:10.1214/09-SS054 (2010). - DOI

Publication types

Actions
Actions

MeSH terms

Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions
Actions

Save citation to file

Email citation

Add to Collections

Add to My Bibliography

Your saved search

Create a file for external citation management software

Your RSS Feed

Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses

Affiliations

Oncoprotein-specific molecular interaction maps (SigMaps) for cancer network analyses

Authors

Affiliations

Abstract

Conflict of interest statement

Figures

Similar articles

Cited by

References

Methods-only References

Publication types

MeSH terms

Substances

Grants and funding

LinkOut - more resources

Full Text Sources

Miscellaneous

Abstract

Conflict of interest statement

Figures

Similar articles

Cited by

References

Methods-only References

Publication types

MeSH terms

Substances

Related information

Grants and funding

LinkOut - more resources

Full Text Sources

Miscellaneous