I just start learning bioinformatics from RNA-seq/DEGs. I am trying to replicate an article. I very much appreciate any help. I used salmon in combination with tximport and DESeq2. PCA plot showed 77%PC1 8%PC2. Apparently, different conditions are mixed and not well separated. Is it healthy to continue this analysis? PCA


sample_table <- read.csv("SraTable.txt") %>% arrange(sample_table$Run) %>% select(Run)

sample_table$Run <- c("pre_ischemia1","ischemia1","pre_ischemia2","ischemia2","pre_ischemia3", "ischemia3","pre_ischemia4","ischemia4","pre_ischemia5","ischemia5")

SRR10700832_quant/quant.sf sample_files <- paste0(c("SRR10700832_quant","SRR10700833_quant","SRR10700835_quant","SRR10700836_quant","SRR10700838_quant", "SRR10700839_quant","SRR10700841_quant","SRR10700842_quant","SRR10700844_quant","SRR10700846_quant"), "/quant.sf")

names(sample_files) <- pull(sample_table, Run)
gene_map <- read.csv("gene_map.csv", col.names = c("enstid", "ensgid")) 
count_data <- tximport(files = sample_files, type="salmon", tx2gene= gene_map, ignoreTxVersion = TRUE)
sample_table <- as.data.frame(sample_table)
sample_table$condition <- factor(rep(c("pre_ischemia", "ischemia"), times=5), levels = c("pre_ischemia", "ischemia"))

deseq_dataset <- DESeqDataSetFromTximport(txi= count_data, colData = sample_table, design = ~condition) 
deseq_dataset <- DESeq(deseq_dataset)
vst <- varianceStabilizingTransformation(deseq_dataset)

plotPCA(vst, intgroup='condition')
results <- results(deseq_dataset)


res <- results(deseq_dataset) yielded following;

out of 38464 with nonzero total read count
adjusted p-value < 0.1
LFC > 0 (up)       : 1, 0.0026%
LFC < 0 (down)     : 0, 0%
outliers [1]       : 263, 0.68%
low counts [2]     : 0, 0%
(mean count < 0)
[1] see 'cooksCutoff' argument of ?results
[2] see 'independentFiltering' argument of ?results

There is only 1 gene upregulated, zero down and zero low counts which are not likely. Which step could be the problem ? I would like to know sanity checkpoint in the entire flow.

Thank you for your time

P.S sorry it this hurts your eyes. I couldnt format the code in the post.

Source link