For the following data, and R code,

https://filebin.net/5zlqg5of4c2o37vu

I got only 1948 genes returned, but count_data.tsv has 26506 rows. Am I supposed to get so few genes with non NA padj. Is there anything wrong with the DESeq2 code that I use? Thanks.

R> sum(!is.na(res$padj))
[1] 1948

count_file='count_data.tsv'
sample_file='meta_data.tsv'
condA='A'
condB='B'
qthr=0.5

read.tsv = function (file, header = T, sep = "\t", quote = "", check.names = F, comment.char = "", stringsAsFactors = F, ...) {
    read.table(file, header = header, sep = sep, quote = quote, check.names = check.names, comment.char = comment.char, stringsAsFactors = stringsAsFactors, ...)
}

sample_table=read.tsv(sample_file)
count_table=read.tsv(count_file)[, sample_table[, 1], drop=F]
design=as.formula(sprintf('~%s', names(sample_table)[[2]]))

suppressWarnings(suppressPackageStartupMessages(library(DESeq2)))
scaled_count_table=counts(
    estimateSizeFactors(
        suppressWarnings(
            DESeqDataSetFromMatrix(
                countData=count_table
                , colData=sample_table[, 2, drop=F]
                , design=design
                )
            )
        )
    , normalized=T
    )
rs=rowSums(scaled_count_table)
sub_count_table=count_table[rs > quantile(rs, qthr), , drop=F]

cds=estimateSizeFactors(
    suppressWarnings(
        DESeqDataSetFromMatrix(
            countData=sub_count_table
            , colData=sample_table
            , design=design
            )
        )
    )

res=data.frame(
    results(
        DESeq(cds, quiet = T)
        , contrast=c(names(sample_table)[[2]], condA, condB)
        )
    )

You seem to be making your life complicated with this. For a simple A vs B comparison, the Quick start section of the vignette should suffice.

To understand NA p-values, please see this section Note on p-values set to NA

Kevin