TP53 PPI + GNN Dashboard

Unified output view for the TP53 protein interaction pipeline, combining network topology, source overlap, evidence distribution, top interactors, and downstream GNN model performance.

2757Total Interactors
501BioGRID + STRING Overlap
265STRING Only
1991BioGRID Only
GCNBest Model
0.8714Best Test Accuracy

TP53 PPI Subgraph

Focused view of the highest-signal first-hop TP53 neighborhood, with interactor-interactor edges preserved for graph learning.

91Displayed Nodes
2507Displayed Edges
90STRING-Supported Nodes
90Max Interactors Shown
TP53 seed Direct STRING support BioGRID-only direct support
99.8%99.9%99.9%99.9%99.9%99.7%99.8%99.9%99.7%99.9%99.9%99.9% TP53MYCRPA1RPA2EGFRESR1HSPA8BRD4TRIM25EP300BRCA1UBCHSPA1APARP1NPM1HSPA5WDR5TRIM28ELAVL1HSP90AA1VCPYAP1CEBPACTNNB1HDAC1XPO1FBXW7HUWE1COPS5CHD4PHBXRCC6ARHSP90AB1MEN1SQSTM1PMLHNRNPUCDK2CUL7ACTBHDAC2PRMT1CYLDPTENCUL1CSNK2A1MOV10RPS6APEX1MDM2RPS27ANCLGAPDHCHD3USP7AURKBCDC42YWHAZRPS3RNF2GSK3BOBSL1SIRT7CDK9G3BP1PLK1DDB1AKT1YWHAEBTRCHSPA4CDK1RPS19PRKDCRPLP0RPL11KDM1ACUL4ARPL5SSRP1EWSR1IFI16NR3C1TUBBAURKABMI1HSPA9SUMO2EEF2SUZ12

Interaction Summary

InteractorSourcesSource CountMax Score
ATMBioGRID,STRING2999
BCL2BioGRID,STRING2999
BCL2L1BioGRID,STRING2999
BRCA1BioGRID,STRING2999
CDKN1ABioGRID,STRING2999
CDKN2ABioGRID,STRING2999
CHEK2BioGRID,STRING2999
CREBBPBioGRID,STRING2999
DAXXBioGRID,STRING2999
EP300BioGRID,STRING2999
HDAC1BioGRID,STRING2999
HIF1ABioGRID,STRING2999
HIPK2BioGRID,STRING2999
HSP90AA1BioGRID,STRING2999
MDM2BioGRID,STRING2999

Evidence Distribution

Evidence TypeCount
Affinity Capture-MS1134
STRING_combined766
Affinity Capture-Western454
Proximity Label-MS225
Negative Genetic141
Reconstituted Complex120
Two-hybrid107
Synthetic Growth Defect96

Direct TP53 Confidence Links

Readable direct-link view in the style of TP53 - 83% - ATM, using the interaction support score as confidence.

TP53 99.9% RPA1
TP53 99.9% EP300
TP53 99.9% BRCA1
TP53 99.9% HSP90AA1
TP53 99.9% HDAC1
TP53 99.9% PTEN
TP53 99.9% MDM2
TP53 99.9% USP7
TP53 99.8% AURKA
TP53 99.8% HSPA9
TP53 99.7% MYC
TP53 99.7% NPM1
TP53 99.7% HSPA4
TP53 99.7% PRKDC
TP53 99.5% HSP90AB1
TP53 99.4% AKT1
TP53 99.2% ESR1
TP53 99.2% GSK3B

Mutant p53 Reactivation Targets

Prioritized proteins for restoring mutant p53 pathway activity, combining TP53 mutation burden, hotspot distribution, reporter-loss context, and TP53-centered network support.

14Curated Candidates Ranked
49929Pathogenic Missense Records
0.99DNA-Binding Mutant Fraction
0.94Reporter Loss Score

Mutation Context

Hotspot codons: 273:4867, 248:4674, 175:3423, 245:2127, 282:1737

Dominant domains: DNA binding, HCD IV - DNA binding, HCD V - DNA binding

Reporter medians: WAF1 2.7, MDM2 8.7, BAX 7.5, GADD45 3.2

Interpretation

Most pathogenic missense variants cluster in the DNA-binding region (98.7%), so restoring transcriptional output is the main biological objective.

The reporter-loss score is high (0.94), indicating broad collapse of canonical p53 target-gene activity.

The strongest inhibition-side targets are MDM2, MDM4, SIRT1, which points to the MDM2/MDM4 and deacetylase axis as the clearest suppression program.

The top restoration-side targets are EP300, CREBBP, HIPK2, highlighting acetylation and DNA-damage signaling as the most plausible recovery routes.

Mechanism Mix

MechanismCount
acetyltransferase3
dna damage kinase3
negative regulator2
deacetylase2
checkpoint kinase2
#1

MDM2

99.47

inhibit

Suppresses p53 stability and transcriptional activity through ubiquitin-mediated control.

negative regulator BioGRID,STRING
#2

MDM4

98.47

inhibit

Blunts p53 transactivation and cooperates with MDM2 in dampening p53 function.

negative regulator BioGRID,STRING
#3

SIRT1

94.75

inhibit

Removes activating acetylation marks from p53 and weakens transcriptional recovery.

deacetylase BioGRID,STRING
#4

EP300

94.25

activate or support

Acetyltransferase activity can restore transcriptional competence to p53-responsive programs.

acetyltransferase BioGRID,STRING
#5

HDAC1

93.75

inhibit

Chromatin and p53 deacetylation can suppress mutant p53 reactivation programs.

deacetylase BioGRID,STRING
#6

CREBBP

93.25

activate or support

Supports p53-dependent transcription through acetylation and co-activation functions.

acetyltransferase BioGRID,STRING
TargetPriorityActionClassSourcesRationale
MDM299.47inhibitnegative regulatorBioGRID,STRINGSuppresses p53 stability and transcriptional activity through ubiquitin-mediated control.
MDM498.47inhibitnegative regulatorBioGRID,STRINGBlunts p53 transactivation and cooperates with MDM2 in dampening p53 function.
SIRT194.75inhibitdeacetylaseBioGRID,STRINGRemoves activating acetylation marks from p53 and weakens transcriptional recovery.
EP30094.25activate or supportacetyltransferaseBioGRID,STRINGAcetyltransferase activity can restore transcriptional competence to p53-responsive programs.
HDAC193.75inhibitdeacetylaseBioGRID,STRINGChromatin and p53 deacetylation can suppress mutant p53 reactivation programs.
CREBBP93.25activate or supportacetyltransferaseBioGRID,STRINGSupports p53-dependent transcription through acetylation and co-activation functions.
HIPK292.61activate or supportactivating kinaseBioGRID,STRINGHIPK2 promotes activating p53 phosphorylation and apoptotic gene expression.
KAT592.21activate or supportacetyltransferaseBioGRID,STRINGTIP60/KAT5-mediated acetylation is linked to p53 activation and apoptotic recovery.
USP791.74inhibitdeubiquitinase axisBioGRID,STRINGCan stabilize the MDM2 axis and indirectly restrain p53 reactivation.
ATM91.03activate or supportdna damage kinaseBioGRID,STRINGDNA-damage signaling can reactivate p53 pathway output through phosphorylation and checkpoint control.

GNN Output

GCN

reasonable_fit
0.8714Test Accuracy
0.8365Balanced Accuracy
0.7625F1 Score

Train accuracy 0.8446, test loss 0.3758. Detailed dashboard: n/a

RANDOM_FOREST

reasonable_fit
0.8228Test Accuracy
0.7974Balanced Accuracy
0.6994F1 Score
No epoch curve for this model.

Train accuracy 0.8512, test loss 0.3881. Detailed dashboard: tp53_random_forest_metrics.html

XGBOOST

reasonable_fit
0.8427Test Accuracy
0.8172Balanced Accuracy
0.7290F1 Score
No epoch curve for this model.

Train accuracy 0.9080, test loss 0.3503. Detailed dashboard: tp53_xgboost_metrics.html

ENSEMBLE

reasonable_fit
0.8644Test Accuracy
0.8023Balanced Accuracy
0.7312F1 Score
No epoch curve for this model.

Train accuracy 0.8996, test loss 0.3640. Detailed dashboard: tp53_ensemble_metrics.html

ModelTest AccBal AccPrecisionRecallSpecificityF1TP/TN/FP/FN
GCN0.87140.83650.76510.76000.91290.7625114/367/35/36
RANDOM_FOREST0.82280.79740.66280.74030.85460.6994114/341/58/40
XGBOOST0.84270.81720.70060.75970.87470.7290117/349/50/37
ENSEMBLE0.86440.80230.81600.66230.94240.7312102/376/23/52

10 Held-Out Prediction Examples

These are unseen test-split examples from ENSEMBLE. Use them as a reviewer-friendly case study to explain why the model is making sensible calls protein by protein.

10Examples Shown
7Correct In Demo
0.8644Full Test Accuracy
0.7312Full Test F1

Downloadable CSV: tp53_ensemble_demo_predictions.csv

ProteinTrue LabelPredictedProbabilityOutcomeDirect BioGRIDBioGRID DegreeDistance to TP53Experiment Diversity
RPA1STRING-supportedSTRING-supported0.8344Correct11145113
ESR1STRING-supportedSTRING-supported0.8348Correct1878113
ESR2STRING-supportedSTRING-supported0.9812Correct070628
HDAC1STRING-supportedSTRING-supported0.8818Correct1508114
USP7STRING-supportedSTRING-supported0.9115Correct1401114
LARP7BioGRID-onlyBioGRID-only0.6027Correct148519
UFL1BioGRID-onlyBioGRID-only0.4446Correct147919
CUL3BioGRID-onlySTRING-supported0.8015Mismatch1998113
EFTUD2BioGRID-onlySTRING-supported0.6453Mismatch1668112
RBM39BioGRID-onlySTRING-supported0.7846Mismatch1513114

Single-Sample Upload Demo

Upload one of the prepared one-row CSV files from your local computer. The page will validate it against the saved ENSEMBLE held-out examples and show the prediction immediately. In this dataset, negative samples mean BioGRID-only (not STRING-supported).

Use one of the files below. This static browser demo validates prepared held-out samples one at a time.

S01: RPA1 BioGRID + STRING BioGRID + STRING S02: ESR1 BioGRID + STRING BioGRID + STRING S03: ESR2 STRING-only STRING-only S04: HDAC1 BioGRID + STRING BioGRID + STRING S05: USP7 BioGRID + STRING BioGRID + STRING S06: LARP7 BioGRID-only (not STRING-supported) Correct negative example S07: UFL1 BioGRID-only (not STRING-supported) Correct negative example S08: CUL3 BioGRID-only (not STRING-supported) BioGRID-only (not STRING-supported) S09: EFTUD2 BioGRID-only (not STRING-supported) BioGRID-only (not STRING-supported) S10: RBM39 BioGRID-only (not STRING-supported) BioGRID-only (not STRING-supported)
No sample uploaded yet.

Recommended files to demonstrate correct negative predictions: S06_ufl1.csv and S07_mepce.csv.