Merge pull request #50 from shahcompbio/develop

Merge changes to cohort overview with updated patient inclusion list

Author: ivazquez

110_cohort_overview.Rmd

@@ -218,26 +218,31 @@ snv_cna_fusions_tbl <- bind_rows(snv_tbl, cna_tbl, fusions_tbl) %>%
 ```{r}
 # Remove scRNA samples from unidentified sites
 db$sequencing_scrna <- db$sequencing_scrna %>% 
+  filter(patient_id %in% included_patients) %>%
   filter(tumor_type != "Unknown") %>%
   filter(therapy == "pre-Rx")
 
 # Remove H&E samples that are not adjacent or site-matched slides
 db$he_slide <- db$he_slide %>% 
+  filter(patient_id %in% included_patients) %>%
   filter(is_adjacent == TRUE) %>%
   filter(therapy == "pre-Rx")
 
 # Remove mpIF samples that have not been delivered
 db$mpif_slide <- db$mpif_slide %>% 
+  filter(patient_id %in% included_patients) %>%
   filter(submission_status == "Delivered") %>%
   filter(therapy == "pre-Rx")
 
 # Remove normal WGS samples and failed samples
 db$sequencing_bulk_dna <- db$sequencing_bulk_dna %>% 
+  filter(patient_id %in% included_patients) %>%
   filter(tumor_type %in% c("Primary","Metastasis")) %>% 
   filter(qc_status == "Pass")
 
 # Remove normal IMPACT samples and remove duplicates
 db$sequencing_msk_impact_custom <- db$sequencing_msk_impact_custom %>% 
+  filter(patient_id %in% included_patients) %>%
   left_join(samples %>% dplyr::select(impact_dmp_sample_id, tumor_purity)) %>%
   filter(impact_gene_panel == "IMPACT") %>%
   group_by(impact_dmp_patient_id) %>%
@@ -246,6 +251,7 @@ db$sequencing_msk_impact_custom <- db$sequencing_msk_impact_custom %>%
 
 # Remove Myriad samples not yet delivered
 db$sequencing_myriad <- db$sequencing_myriad %>% 
+  filter(patient_id %in% included_patients) %>%
   filter(status == "Delivered")
 
 patients <- db$patients %>%
@@ -363,7 +369,7 @@ mat <- data %>%
 ```{r, fig.width = 8, fig.height = 4.5}
 
 snv_cna_fusions_wide_tbl <- snv_cna_fusions_tbl %>%
-  filter(patient_id %in% scrna_patients) %>%
+  filter(patient_id %in% Reduce(union, list(scrna_patients, hne_patients, mpif_patients))) %>%
   transform(patient_id = str_remove_all(patient_id, "SPECTRUM-OV-")) %>%
   pivot_wider(
     id_cols = "hugo_symbol",
@@ -397,7 +403,7 @@ oncoprint_main_mat <- snv_cna_fusions_wide_tbl %>%
 
 # Top annotation
 top_df <- db$mutational_signatures %>%
-  filter(patient_id %in% scrna_patients) %>%
+  filter(patient_id %in% Reduce(union, list(scrna_patients, hne_patients, mpif_patients))) %>%
   left_join(patients, by = 'patient_id') %>%
   left_join(gyn_diagnosis, by = 'patient_id') %>%
   mutate(patient_id = str_remove_all(patient_id, "SPECTRUM-OV-")) %>%
@@ -479,11 +485,11 @@ rownames(oncoprint_main_ht@matrix)
 
 colnames(oncoprint_main_ht@matrix)
 
-cairo_pdf("figures/110_cohort_overview/110_oncoprint_spectrum_scrna_patients.pdf", width = 13.5, height = 4.5)
+cairo_pdf("figures/110_cohort_overview/110_oncoprint_spectrum_scrna_hne_mpif_patients.pdf", width = 13.5, height = 4.5)
 print(oncoprint_main_ht)
 dev.off()
 
-png("figures/110_cohort_overview/110_oncoprint_spectrum_scrna_patients.png", width = 13.5, height = 4.5, res = 300)
+png("figures/110_cohort_overview/110_oncoprint_spectrum_scrna_hne_mpif_patients.png", width = 13.5, height = 4.5, res = 300)
 print(oncoprint_main_ht)
 dev.off()
 
@@ -614,7 +620,7 @@ dev.off()
 ```{r}
 
 data <- inventory %>%
-  filter(patient_id %in% scrna_patients) %>%
+  filter(patient_id %in% Reduce(union, list(scrna_patients, hne_patients, mpif_patients))) %>%
   filter((therapy == "pre-Rx") | (therapy == "post-Rx" & technique == "bulk_dna")) %>%
   get_sample_inventory(., unique = TRUE) %>% 
   dplyr::left_join(patients, by = 'patient_id') %>%

figures/110_cohort_overview/110_cohort_overview.pdf

@@ -112,16 +112,34 @@ included_patients <- db$patients %>%
 
 # Define patients included in the study with scRNA data
 scrna_patients <- db$sequencing_scrna %>%
-  filter(patient_id %in% included_patients) %>%
-  filter(therapy == "pre-Rx") %>%
-  filter(platform == "10x 3' GE") %>%
+  filter(patient_id %in% included_patients,
+         therapy == "pre-Rx",
+         platform == "10x 3' GE",
+         qc_status == "Pass") %>%
+  pull(patient_id) %>%
+  unique
+
+# Define patients included in the study with mpIF data
+mpif_patients <- db$mpif_slide %>%
+  filter(patient_id %in% included_patients,
+         therapy == "pre-Rx",
+         qc_status == "Pass") %>%
+  pull(patient_id) %>%
+  unique
+
+# Define patients included in the study with H&E data
+hne_patients <- db$he_slide %>%
+  filter(patient_id %in% included_patients,
+         patient_id %in% union(scrna_patients, mpif_patients),
+         therapy == "pre-Rx") %>%
   pull(patient_id) %>%
   unique
 
 ## load mutational signatures ----------------------
 
 signature_tbl <- db$mutational_signatures %>%
-  mutate(consensus_signature = ordered(consensus_signature, levels = names(clrs$consensus_signature))) %>% 
+  mutate(consensus_signature = ordered(consensus_signature, levels = names(clrs$consensus_signature)),
+         consensus_signature_short = ordered(consensus_signature_short, levels = names(clrs$consensus_signature_short))) %>% 
   arrange(patient_id)
 
 ## load scRNA meta data -----------------------------
@@ -136,6 +154,13 @@ scrna_meta_tbl <- db$sequencing_scrna %>%
          tumor_supersite = str_replace_all(tumor_supersite, "Upper Quadrant", "UQ")) %>% 
   mutate(tumor_megasite = ifelse(!tumor_supersite %in% c("Adnexa", "Ascites"),
                                  "Other", tumor_supersite)) %>% 
+  mutate(tumor_megasite_adnexa = case_when(
+    tumor_megasite == "Adnexa" ~ "Adnexa",
+    tumor_megasite == "Ascites" ~ "Other",
+    tumor_megasite == "Other" ~ "Other",
+    tumor_megasite == "Unknown" ~ "Unknown",
+    TRUE ~ as.character(tumor_megasite)
+  )) %>%
   mutate(tumor_supersite = ordered(tumor_supersite, levels = names(clrs$tumor_supersite))) %>%
   mutate(
     tumor_supersite_adnexa = case_when(
@@ -147,20 +172,6 @@ scrna_meta_tbl <- db$sequencing_scrna %>%
   ) %>%
   left_join(signature_tbl, by = "patient_id")
 
-## load mpIF meta data -------------------------------
-
-mpif_slide_meta_tbl <- db$mpif_slide %>%
-  mutate(slide_id = str_replace_all(pici_id, " ", "_"),
-         sample_id = paste0(patient_id, "_", surgery, str_replace_all(toupper(tumor_subsite), " ", "_"))) %>%
-  mutate(patient_id_short = str_remove_all(patient_id, "SPECTRUM-OV-"),
-         tumor_supersite = str_replace_all(tumor_supersite, "Upper Quadrant", "UQ")) %>% 
-  mutate(tumor_megasite = ifelse(!tumor_supersite %in% c("Adnexa"),
-                                 "Other", tumor_supersite)) %>% 
-  mutate(tumor_supersite = ordered(tumor_supersite, levels = names(clrs$tumor_supersite))) %>% 
-  filter(patient_id %in% included_patients,
-         therapy == "pre-Rx") %>% 
-  left_join(signature_tbl, by = "patient_id")
-
 ## load H&E meta data -------------------------------
 
 hne_meta_tbl <- db$he_slide %>%
@@ -169,11 +180,60 @@ hne_meta_tbl <- db$he_slide %>%
          tumor_supersite = str_replace_all(tumor_supersite, "Upper Quadrant", "UQ")) %>%
   mutate(tumor_megasite = ifelse(!tumor_supersite %in% c("Adnexa", "Ascites"),
                                  "Other", tumor_supersite)) %>%
+  mutate(tumor_megasite_adnexa = case_when(
+    tumor_megasite == "Adnexa" ~ "Adnexa",
+    tumor_megasite == "Ascites" ~ "Other",
+    tumor_megasite == "Other" ~ "Other",
+    tumor_megasite == "Unknown" ~ "Unknown",
+    TRUE ~ as.character(tumor_megasite)
+  )) %>%
   mutate(tumor_supersite = ordered(tumor_supersite, levels = names(clrs$tumor_supersite))) %>%
+  mutate(
+    tumor_supersite_adnexa = case_when(
+      tumor_supersite == "Adnexa" & grepl("Adnexa", tumor_subsite) ~ "Adnexa",
+      tumor_supersite == "Adnexa" & grepl("Ovary", tumor_subsite) ~ "Ovary",
+      tumor_supersite == "Adnexa" & grepl("Fallopian Tube", tumor_subsite) ~ "Fallopian Tube",
+      TRUE ~ as.character(tumor_supersite)
+    )
+  ) %>%
   filter(patient_id %in% included_patients,
          therapy == "pre-Rx") %>%
   left_join(signature_tbl, by = "patient_id")
 
+## load mpIF meta data -------------------------------
+
+mpif_slide_meta_tbl <- db$mpif_slide %>%
+  # mutate(slide_id = str_replace_all(pici_id, " ", "_"),
+  mutate(slide_id = paste0(patient_id, "_", surgery, str_replace_all(toupper(tumor_subsite), " ", "_")),
+         patient_id_short = str_remove_all(patient_id, "SPECTRUM-OV-"),
+         sample_id_short = str_remove_all(sample_id, "OV-"),
+         aliquot_id_short = str_remove_all(aliquot_id, "OV-"),
+         tumor_supersite = str_replace_all(tumor_supersite, "Upper Quadrant", "UQ")) %>% 
+  mutate(tumor_megasite = ifelse(!tumor_supersite %in% c("Adnexa"),
+                                 "Other", tumor_supersite)) %>% 
+  mutate(tumor_megasite_adnexa = case_when(
+    tumor_megasite == "Adnexa" ~ "Adnexa",
+    tumor_megasite == "Ascites" ~ "Other",
+    tumor_megasite == "Other" ~ "Other",
+    tumor_megasite == "Unknown" ~ "Unknown",
+    TRUE ~ as.character(tumor_megasite)
+  )) %>%
+  mutate(tumor_supersite = ordered(tumor_supersite, levels = names(clrs$tumor_supersite))) %>% 
+  mutate(
+    tumor_supersite_adnexa = case_when(
+      tumor_supersite == "Adnexa" & grepl("Adnexa", tumor_subsite) ~ "Adnexa",
+      tumor_supersite == "Adnexa" & grepl("Ovary", tumor_subsite) ~ "Ovary",
+      tumor_supersite == "Adnexa" & grepl("Fallopian Tube", tumor_subsite) ~ "Fallopian Tube",
+      TRUE ~ as.character(tumor_supersite)
+    )
+  ) %>%
+  filter(patient_id %in% included_patients,
+         therapy == "pre-Rx",
+         qc_status == "Pass") %>% 
+  left_join(signature_tbl, by = "patient_id")
+
+mpif_fov_meta_tbl <- db$mpif_fov
+
 ## load WGS meta data -------------------------------
 
 bulk_dna_meta_tbl <- db$sequencing_bulk_dna %>%