Discovering Dominant Tumor Immune Archetypes in a Pan-Cancer Census

This repository contains code used to create figures for this paper All RNAseq and single cell data is available using the GEO accession GSE184398. Some data that is not available on GEO or UCSF Data Library has been added to this folder files_used_for_plots

Analysis

Figure 1

1. Figure 1C get_flow_score.py input files are in files_used_for_plots

2. Figure 1D correlation_score_from_flow.py

Input files should have this format (example file for Tcell flow vs Score correlation `files_used_for_plots/tcell_flow_and_Score.tsv`). 
	

				|flow_Tcell |Score
	IPIBLAD032.T1.rna.live	|  31.3	    |  46.98

	Input files can be made from calculating scores with get_score.py and joining the corresponding flow data

`python correlation_score_from_flow.py ../files_used_for_plots/tcell_flow_and_Score.tsv`

1. Figure 1E cross_whisker_plots_flow.py

   input files needed

    	1. feature scores tsv file output of `get_score.py` (example file in `/files_used_for_plots/tcell_score_percentile_of_percentiles.tsv`
    	2. flow scores tsv file output of `get_flow_score.py` (example file in `/files_used_for_plots/Flow_score_Sept10_2020_Tcell_percentile_of_percent.tsv`
   

Figure 2

1. Figures 2A-C, 2E-F, 2J, 2L 3_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI. 
    		2. The feature scores for TCGA using not included but can be generated using TCGA data and `get_score.py``
   

2. Figures 2H-I, 2K boxplots made using seaborn.boxplot with data in /files_used_for_plots/feature_matrix.txt

3. Figure 2M heatmap was made using seaborn.clustermap with medians calculated from TPMs in /files_used_for_plots/feature_matrix.txt

4. Figure 2N barplot and bubble plots were made using seaborn.barplot and seaborn.scatter respectively with medians calculated from TPMs in ``/files_used_for_plots/feature_matrix.txt` for IPI the chemokine TPMs for TCGA are not included

Figure 3

1. Figures 3A 3_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI. 
   

2. Figures 3B-C TCGA_survival_3_features.py

   input files needed

   Can be found here XENA Browser

    	   1. A folder with TCGA clinical data files for each of these indications ('HNSC','KIRC','SKCM','BLCA','SARC','OV','UCEC','UCS','COAD','LIHC','LUAD','PAAD','GBM') 			   
    	   
    	    TCGA
    			|_ BLCA
    			|	|_ BLCA_clinical_data.tsv
    			|
    			|_COAD
    			|	|_ COAD_clinical_data.tsv
    			.
    			.
    			.
    	   
    	   2. TCGA archetypes file `/files_used_for_plots/TCGA_archetypes.tsv`
   

3. Figures 3D-E boxplots made using seaborn.boxplot with data in /files_used_for_plots/feature_matrix.txt

Figure 4

1. Figure 4A get_flow_score.py

2. Figure 4B heatmap was made using seaborn.clustermap

3. Figures 4C 6_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI.
   

4. Figures 4D,4G boxplots made using seaborn.boxplot with data in /files_used_for_plots/feature_matrix.txt

5. Figures 4E the alluvial plot was made using RAWgraphs 2.0

6. Figure 4F heatmap was made using seaborn.clustermap with medians calculated from TPMs in /files_used_for_plots/feature_matrix.txt

Figure 5

1. Figure 5A get_flow_score.py

2. Figures 5A, 5O, 5R 10_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI.
   

3. Figures 5B the alluvial plot was made using RAWgraphs 2.0

4. Figures 5D-N boxplots and scatter plots were made using seaborn.boxplot and seaborn.scatter respectively.

5. Figures 5P-Q, 5S-T boxplots made using seaborn.boxplot with data in /files_used_for_plots/feature_matrix.txt

Figure 6

1. Figures 6A-D heatmap and bubble plots were made using seaborn.clustermap and seaborn.scatter respectively with medians of chemokines calculated from TPMs in ``/files_used_for_plots/feature_matrix.txt`. TPMS for other genes can be found on using the GEO accession GSE184398

Figure 7

1. Figures 7A-B 10_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI.
   

2. Figures 7B-E boxplot, heatmaps and bubble plot were made using seaborn.boxplot seaborn.clustermap and seaborn.scatter respectively.TPMS for genes can be found on using the GEO accession GSE184398

3. Figures 7F TCGA_survival_10_features.py

   input files needed

   Can be found here XENA Browser

    	   1. A folder with TCGA clinical data files for each of these indications ('HNSC','KIRC','SKCM','BLCA','SARC','OV','UCEC','UCS','COAD','LIHC','LUAD','PAAD','GBM') 			   
    	   
    	    TCGA
    			|_ BLCA
    			|	|_ BLCA_clinical_data.tsv
    			|
    			|_COAD
    			|	|_ COAD_clinical_data.tsv
    			.
    			.
    			.
    	   
    	   2. TCGA archetypes file `/files_used_for_plots/TCGA_archetypes_assigned.tsv` 				
   

##Supplementary Figures

Figure 1S

1. Figure 1SA

   Kmeans clustering using sklearn.cluster.KMeans of and dataframe of TPMS of all compartments. These can be found on using the GEO accession GSE184398

2. Figure 1SB

    makes a volcano plot from a standard Limma DGE output
    e.g.
               logFC    |   AveExpr	    |       t	    |  P.Value	|  adj.P.Val	|      B
     PLK1	7.374675473	| 1.415522735	| 6.886987802	| 3.92E-06	| 0.050802876	| 3.971754014
     ZWINT	5.772193613	| 1.340166868	| 5.872941797	| 2.48E-05	| 0.065626979	| 2.450364751
    		
    
    python volcano_plots.py -h
    
    Usage: volcano_plots.py -i <input file name> [options]
    
    Make Volcano Plots
    
    Options:
      -h, --help      show this help message and exit
      -i <input file>   Path and name of input file
      -o <output file>  Path and names of output files e.g DGE_treg_vs_tcell (a .png,.pdf & .svg file extension will automatically be added)
      -p <p-value>      P-value cutoff DEFAULT 0.005           
      -t <title>      Title in quotes e.g. "Tregs vs Tcells"
   

3. Figure 1SC correlation_score_from_flow.py see figure 1D

4. Figure 1SD get_score.py python get_score.py -h

    Usage: get_score.py -o <output file name> -g <gene signature file name> -f <gene expresson tsv>  [options] ALL OPTIONS ARE REQUIRED
    
    Get gene signature score
    
    Options:
      -h, --help      show this help message and exit
      -t <title>      title used for all output files
      -g <gene sig>   path to a file with a list of genes (HUGO Names) in gene
                      signature.
      -f <gene file>  path to normalized gene expression file (TPM or logCPM) with
                      samples in the columns and genes in the rows 
   

5. Figure 1SE cross_whisker_plots_flow.py see figure 1E

Figure 2S

1. Figure 2SA & 2SB DBI_frequency_of_clusters.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI.
   

2. Figures 2SD 3_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt for IPI.
   

3. Figures 2SE-F boxplots and scatter plots were made using seaborn.boxplot and seaborn.scatter respectively.

Figure 3S

1. Figures 3SA 3_feature_UMAP_IPI.py

2. Figures 3SB-D TCGA_survival_3_features.py

   input files needed

   Can be found here XENA Browser

    	   1. A folder with TCGA clinical data files for each of these indications ('HNSC','KIRC','SKCM','BLCA','SARC','OV','UCEC','UCS','COAD','LIHC','LUAD','PAAD','GBM') 			   
    	   
    	    TCGA
    			|_ BLCA
    			|	|_ BLCA_clinical_data.tsv
    			|
    			|_COAD
    			|	|_ COAD_clinical_data.tsv
    			.
    			.
    			.
    	   
    	   2. TCGA archetypes file `/files_used_for_plots/TCGA_archetypes.tsv`      
   

Figure 4S

1. Figure 4SB

    makes a volcano plot from a standard Limma DGE output
    e.g.
               logFC    |   AveExpr	    |       t	    |  P.Value	|  adj.P.Val	|      B
     PLK1	7.374675473	| 1.415522735	| 6.886987802	| 3.92E-06	| 0.050802876	| 3.971754014
     ZWINT	5.772193613	| 1.340166868	| 5.872941797	| 2.48E-05	| 0.065626979	| 2.450364751
    		
    
    python volcano_plots.py -h
    
    Usage: volcano_plots.py -i <input file name> [options]
    
    Make Volcano Plots
    
    Options:
      -h, --help      show this help message and exit
      -i <input file>   Path and name of input file
      -o <output file>  Path and names of output files e.g DGE_treg_vs_tcell (a .png,.pdf & .svg file extension will automatically be added)
      -p <p-value>      P-value cutoff DEFAULT 0.005           
      -t <title>      Title in quotes e.g. "Tregs vs Tcells"
   

2. Figure 4SC, 4Sk correlation_score_from_flow.py see figure 1D

3. Figures 4SD, 4SF-G, 4SL, 4SP 6_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt
   

4. Figure 4SE DBI_frequency_of_clusters.py

   input file needed

   		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt.
   

5. Figures 4SG-H, 4SI, 4SN-N boxplots and scatter plots were made using seaborn.boxplot and seaborn.scatter respectively.

Figure 5S

1. Figure 5SF correlation_score_from_flow.py see figure 1D

2. Figure 5SG get_flow_score.py see figure 1C

3. Figure 4SE DBI_frequency_of_clusters.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt.
   

4. Figures 5SI-J 10_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt
   

Figure 6S

1. Figures 6SA, 6SE-F 10_feature_UMAP_IPI.py

   input file needed

    		1. feature matrix tsv file output (example file in `/files_used_for_plots/feature_matrix.txt
   

2. Figures 6SC-E, 6SG boxplots made using seaborn.boxplot with data in /files_used_for_plots/feature_matrix.txt

Figure 7S

1. Figures 7SA-K TCGA_survival_10_features.py

   input files needed

   Can be found here XENA Browser

    	   1. A folder with TCGA clinical data files for each of these indications ('HNSC','KIRC','SKCM','BLCA','SARC','OV','UCEC','UCS','COAD','LIHC','LUAD','PAAD','GBM') 			   
    	   
    	    TCGA
    			|_ BLCA
    			|	|_ BLCA_clinical_data.tsv
    			|
    			|_COAD
    			|	|_ COAD_clinical_data.tsv
    			.
    			.
    			.
    	   
    	   2. TCGA archetypes file `/files_used_for_plots/TCGA_archetypes_assigned.tsv`
   

Contact Info

Bushra Samad (email: Bushra[dot]Samad[at]ucsf[dot]edu) Alexis Combes (email: Alexis[dot]Combes[at]ucsf[dot]edu) Matthew Krummel (email: Matthew[dot]Krummel[at]ucsf[dot]edu)