S. cerevisiae differential expression, merged sample edition.

index.html 01_annotation.html 02_sample_estimation.html

1 Differential Expression: 20170915

2 Filter the raw data

In sample_estimation, I created sc_filt which is precisely what I want.

3 Start with batch in the model

keepers <- list(
  "mcwu_vs_wcwu" = c("mtc_wtu", "wtc_wtu"),
  "wcmu_vs_wcwu" = c("wtc_mtu", "wtc_wtu"),
  "mcmu_vs_mcwu" = c("mtc_mtu", "mtc_wtu"),
  "mcmu_vs_wcmu" = c("mtc_mtu", "wtc_mtu"),
  "mcmu_vs_wcwu" = c("mtc_mtu", "wtc_wtu"))

fsva <- sm(all_pairwise(input=merged_filt, model_batch="fsva", parallel=FALSE))

fsva_write <- sm(combine_de_tables(
  all_pairwise_result=fsva, keepers=keepers,
  excel=paste0("excel/sva_in_model_differential_merged-v", ver, ".xlsx"),
  sig_excel=paste0("excel/sva_in_model_significant-v", ver, ".xlsx"),
  abundant_excel=paste0("excel/sva_in_model_abundance-v", ver, ".xlsx")))

strict_sig_write <- sm(extract_significant_genes(
  fsva_write, fc=2,
  excel=paste0("excel/sva_in_model_sig2lfc-v", ver, ".xlsx")))

3.1 Perform pairwise without the batch ‘r’

merged_nor <- subset_expt(merged_filt, subset="batch!='r'")
fsva_nor <- sm(all_pairwise(input=merged_nor, model_batch="fsva", parallel=FALSE))

fsva_nor_write <- sm(combine_de_tables(
  all_pairwise_result=fsva_nor, keepers=keepers,
  excel=paste0("excel/nor_sva_in_model_differential_merged-v", ver, ".xlsx"),
  abundant_excel=paste0("excel/nor_sva_in_model_abundance-v", ver, ".xlsx"),
  sig_excel=paste0("excel/nor_sva_in_model_significant-v", ver, ".xlsx")))

strict_fsva_nor_write <- sm(extract_significant_genes(
  fsva_nor_write, fc=2,
  excel=paste0("excel/nor_sva_in_model_sig2lfc-v", ver, ".xlsx")))

merged_nos <- subset_expt(merged_filt, subset="batch!='s'")
fsva_nos <- sm(all_pairwise(input=merged_nos, model_batch="fsva", parallel=FALSE))

fsva_nos_write <- sm(combine_de_tables(
  all_pairwise_result=fsva_nos, keepers=keepers,
  excel=paste0("excel/nos_sva_in_model_differential_merged-v", ver, ".xlsx"),
  abundant_excel=paste0("excel/nos_sva_in_model_abundance-v", ver, ".xlsx"),
  sig_excel=paste0("excel/nos_sva_in_model_significant_merged-v", ver, ".xlsx")))

strict_fsva_nos_write <- sm(extract_significant_genes(
  fsva_nor_write, fc=2,
  excel=paste0("excel/nos_sva_in_model_sig2lfs-v", ver, ".xlsx")))

v1_env <- new.env()
old_ver <- "20170515"
load(paste0("../scerevisiae_cbf5v1/savefiles/differential_expression-v", old_ver, ".rda.xz"),
     envir=v1_env)
v1_de <- v1_env$batch_write$data$mtc_vs_wtc
##v1_de <- nobatch_write$data$wtc_wtu_vs_mtc_wtu
##rm(list=c("v1_env"))

merged_de <- fsva$deseq$all_tables$wtc_wtu_vs_mtc_wtu
merged_de$logFC <- merged_de$logFC * -1.0
v1v2_merged <- merge(v1_de, merged_de, by="row.names")
v1v2_merged <- v1v2_merged[, c("deseq_logfc", "logFC")]
v1v2_scatter <- plot_linear_scatter(df=v1v2_merged, loess=TRUE)

## Used Bon Ferroni corrected t test(s) between columns.

v1v2_scatter$scatter

v1v2_scatter$correlation

## 
##  Pearson's product-moment correlation
## 
## data:  df[, 1] and df[, 2]
## t = 46, df = 5800, p-value <2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.5013 0.5389
## sample estimates:
##    cor 
## 0.5204

cor.test(v1v2_merged[, 1], v1v2_merged[, 2], method="spearman")

## Warning in cor.test.default(v1v2_merged[, 1], v1v2_merged[, 2], method = "spearman"):
## Cannot compute exact p-value with ties

## 
##  Spearman's rank correlation rho
## 
## data:  v1v2_merged[, 1] and v1v2_merged[, 2]
## S = 1.7e+10, p-value <2e-16
## alternative hypothesis: true rho is not equal to 0
## sample estimates:
##    rho 
## 0.4798

## Well at least the direction is correct...

merged_nor_de <- fsva_nor$deseq$all_tables$wtc_wtu_vs_mtc_wtu
merged_nor_de$logFC <- merged_nor_de$logFC * -1.0
v1v2nor_merged <- merge(v1_de, merged_nor_de, by="row.names")
v1v2nor_merged <- v1v2nor_merged[, c("deseq_logfc", "logFC")]
v1v2nor_scatter <- plot_linear_scatter(df=v1v2nor_merged, loess=TRUE)

## Used Bon Ferroni corrected t test(s) between columns.

v1v2nor_scatter$scatter

v1v2nor_scatter$correlation

## 
##  Pearson's product-moment correlation
## 
## data:  df[, 1] and df[, 2]
## t = 140, df = 5800, p-value <2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.8684 0.8805
## sample estimates:
##    cor 
## 0.8746

cor.test(v1v2_merged[, 1], v1v2_merged[, 2])

## 
##  Pearson's product-moment correlation
## 
## data:  v1v2_merged[, 1] and v1v2_merged[, 2]
## t = 46, df = 5800, p-value <2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.5013 0.5389
## sample estimates:
##    cor 
## 0.5204

## Well at least the direction is correct...

merged_nos_de <- fsva_nos$deseq$all_tables$wtc_wtu_vs_mtc_wtu
merged_nos_de$logFC <- merged_nos_de$logFC * -1.0
v1v2_merged <- merge(v1_de, merged_nos_de, by="row.names")
v1v2_merged <- v1v2_merged[, c("deseq_logfc", "logFC")]
v1v2_scatter <- plot_linear_scatter(df=v1v2_merged, loess=TRUE)

## Used Bon Ferroni corrected t test(s) between columns.

v1v2_scatter$scatter

v1v2_scatter$correlation

## 
##  Pearson's product-moment correlation
## 
## data:  df[, 1] and df[, 2]
## t = 43, df = 5800, p-value <2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.4680 0.5073
## sample estimates:
##    cor 
## 0.4879

cor.test(v1v2_merged[, 1], v1v2_merged[, 2])

## 
##  Pearson's product-moment correlation
## 
## data:  v1v2_merged[, 1] and v1v2_merged[, 2]
## t = 43, df = 5800, p-value <2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.4680 0.5073
## sample estimates:
##    cor 
## 0.4879

## Well at least the direction is correct...