PRECOG: PREdiction of Clinical Outcomes from Genomic Profiles
PRECOG is a resource for querying associations between genomic profiles and cancer outcomes.
It lets researchers examine whether higher expression of a gene and immune cell abundance
are prognostic for shorter or longer patient survival. PRECOG includes 335 cancer datasets
with survival or response data for >46,000 patients, covering ~55 malignancies.
Gentles/Newman et al.
Nature Medicine (2015)
Nature Medicine (2015)
Benard/Lalgudi et al.
Nucleic Acids Research (2025)
Nucleic Acids Research (2025)
Adult
Explore individual studies, meta-z scores across tumor types, and cell-type associations with outcomes.
Pediatric
Explore individual study and meta-z scores across tumor types for gene and cell type associations with outcomes.
Immunotherapy (ICI)
ICI-focused studies with gene and cell type associations with outcomes across primary and metastatic sites.
Adult PRECOG
Explore Adult PRECOG datasets, meta-Z analysis, and individual-study survival results. This compendium covers non-ICI therapy datasets including The Cancer Genome Atlas (TCGA). Heatmaps visualize the association between a gene’s expression and outcome in a particular dataset, or summarized across datasets. Kaplan-Meier plots can be seen by clicking on a cell in the individual analysis heatmap. Cell-type level associations are also available.
See Gentles/Newman et al. in Nature Medicine (2015) and Benard/Lalgudi et al. in Nucleic Acids Research (2025) for further details.
Data
Individual Analysis
Meta-Z Analysis
Immune Fractions
Immune Fractions Meta-Z
Pediatric PRECOG
Explore Pediatric PRECOG datasets, meta-Z analysis, and individual-study survival results. This compendium covers pediatric cancers only. Heatmaps visualize the association between a gene’s expression and outcome in a particular dataset, or summarized across datasets. Kaplan-Meier plots are currently not available for Pediatric PRECOG. Cell-type level associations are also available.
See Stahl et al. for further details.
Data
Individual Analysis
Meta-Z Analysis
Immune Fractions
Immune Fractions Meta-Z
Immunotherapy (ICI) PRECOG
Explore immune checkpoint inhibitor (ICI) PRECOG datasets, meta-Z analysis, and individual-study survival results. This compendium covers cancers where the primary treatment is ICI. Heatmaps visualize the association between a gene’s expression and outcome in a particular dataset, or summarized across datasets. ICI PRECOG contains a mixture of time-to-event and responder vs non-responder analyses. Kaplan-Meier plots (for survival) or boxplots (for logistic regression of binary response) can be seen by clicking on a cell in the heatmap for individual datasets. Cell-type level associations are also available.
See Benard & Lalgudi et al. in Nucleic Acids Research (2025) for more information.
Data
Individual Analysis
Meta-Z Analysis
Immune Fractions
Immune Fractions Individual Analysis
Immune Fractions Meta-Z
Downloads were updated on 12/07/2025.
If you downloaded data before this date, please re-download the updated files.
This table summarizes data included in ICI PRECOG.
Each row represents a unique dataset for a study based on shared cancer type/subtype,
ICI target, primary/metastatic site, and treatment exposure status.
This table shows survival z-scores collapsed by cancer/cancer suntype.
Filter the table based on search terms (case insensitive) in the search field below.
If you wish to search multiple terms, separate each term with a pipe | (do not add any spaces).
For example, searching with foxm1|klrb1 should yield 2 results.
Survival z-scores collapsed by cancer/cancer subtype are shown.
Filter the table based on search terms (case insensitive) in the search field below.
If you wish to search multiple terms, separate each term with a pipe | (do not add any spaces).
For example, searching with foxm1|klrb1 should yield 2 results.
Survival z-scores for protein coding genes collapsed by cancer type, ICI target, primary/metastatic site, and treatment exposure status are shown.
Use regex search to filter (e.g., foxm1|klrb1). Collapsed z-scores for all genes (including non-protein coding) are available in the downloads tab
The CIBERSORT Signature Matrix (LM22) identifies immune-cell-specific marker genes. Expression levels of these genes are displayed below. A downloadable PCL file is available here.
Inferred fractions of immune cell types can be associated with survival outcomes in a meta-Z analogous to the individual gene meta-Zs found in PRECOG. Cell-type survival z-scores are shown for each collapsed cancer type.
Relative RNA fractions of 22 leukocyte subsets in human cancers
resulting from applying CIBERSORT to select PRECOG and TCGA datasets.
For clarity, values are presented as the mean fraction for each cell
population and cancer type. In accordance with our sensitivity and
specificity analysis, a CIBERSORT p-value threshold of 0.005 was
employed to filter out insignificant deconvolution results.
Relative RNA fractions of 22 leukocyte subsets in human cancers resulting from applying CIBERSORT to select Pediatric PRECOG datasets. For clarity, values are presented as the mean fraction for each cell population and cancer type. In accordance with our sensitivity and specificity analysis, a CIBERSORT p-value threshold of 0.005 was employed to filter out insignificant deconvolution results.
Inferred fractions of immune cell types can be associated with survival outcomes in a meta-Z analogous to the individual gene meta-Zs found in PRECOG. Cell-type survival z-scores are shown for each collapsed cancer type.
Inferred fractions of immune cell types can be associated with survival outcomes in a meta-Z analogous to the individual gene meta-Zs found in PRECOG. Cell-type survival z-scores are shown for each collapsed cancer type.
Relative RNA fractions of 22 leukocyte subsets in human cancers resulting from applying CIBERSORTx to all ICI PRECOG datasets. For clarity, values are presented as the mean fraction for each cell population and cancer type. In accordance with our sensitivity and specificity analysis, a CIBERSORTx p-value threshold of 0.005 was employed to filter out insignificant deconvolution results.
Per-collapsed-group results for CIBERSORTx LM22 immune cell fractions (z-scores), including study-level annotations.
Use the search box (supports regex with
|) to filter rows.Survival z-scores for each individual dataset in PRECOG
are shown. Column headers are datasets that can be cross-referenced in the datasets table (including citation information). Filter the table based on search terms (case insensitive) in the search field below. If you wish to search
multiple terms, separate each term with a pipe | (do not add any spaces).
For example, searching with foxm1|klrb1 should yield 2 results.
Click on the number within any given cell to display the corresponding Kaplan Meier plot(s) for the corresponding gene and dataset(s). KM plots were generated using a median split.
Click on the number within any given cell to display the corresponding Kaplan Meier plot(s) for the corresponding gene and dataset(s). KM plots were generated using a median split.
Survival z-scores for each individua cancer type in TCGA
are shown. Column headers are cancer type abbreviations from TCGA. See more information here. Filter the table based on search terms (case insensitive) in the search field below. If you wish to search
multiple terms, separate each term with a pipe | (do not add any spaces).
For example, searching with foxm1|klrb1 should yield 2 results.
Click on the number within any given cell to display the corresponding Kaplan Meier or Generalized Linear Model plot(s) for the corresponding gene and dataset(s). KM and GLM plots were generated using a median split.
Click on the number within any given cell to display the corresponding Kaplan Meier or Generalized Linear Model plot(s) for the corresponding gene and dataset(s). KM and GLM plots were generated using a median split.
Survival z-scores for protein coding genes in each individual dataset in ICI PRECOG are shown. Use regex search to filter (e.g., foxm1|klrb1). Click any z-score to view Kaplan-Meier or boxplot distribution plots. Individual dataset z-scores for all genes (including non-protein coding) are available in the downloads tab.
Survival z-scores for each individual dataset in Pediatric PRECOG
are shown. Column headers are datasets that can be cross-referenced in the datasets table (including citation information). Filter the table based on search terms (case insensitive) in the search field below. If you wish to search
multiple terms, separate each term with a pipe | (do not add any spaces).
For example, searching with foxm1|klrb1 should yield 2 results
Gene-Gene MetaZ
Plots per-study MetaZ scores: Gene 1 (x) vs Gene 2 (y). Hover a point to see dataset + metadata.
Cell Type × Cell Type (MetaZ)
Plots cohort-collapsed MetaZ scores: Cell 1 (x) vs Cell 2 (y). Hover shows label + metadata.
Cancer–Cancer (MetaZ)
One point per gene; axes are cohort-collapsed meta-Z for the selected cancer types. Hover shows the gene.