The original use case was to concatenate multiple fastqs into a single fastq, while also converting the U's to T's for compatibility with downstream tools. The streaming functionality of FastqStreamer seemed like a good approach to keep the memory usage low while converting each fastq. I could use unix tools ( e.g. cat and awk), but was curious to see how to do it using bioconductor tools.

Your response helped point me in the right direction. I used a BStringSet initially to allow for Us to be converted to Ts, and subsequently coerced to a DNAStringSet. I could then generate a ShortReadQ object and write records to disk with writeFastq. Probably not the most efficient but worked for my initial use case.

Here's my approach, that allows for limiting the # of lines read at a time, in case it is useful to anyone.

suppressPackageStartupMessages(library(ShortRead))

fq <- tempfile()
outfq <- tempfile()
u_fq_txt <- paste(c("@readid", 
                    "UCGA",
                    "+",
                    "]]]]"), 
                  collapse = "\n")
writeLines(u_fq_txt, fq)

convert_rna_fastq <- function(fn, outfn, n_per_chunk = 100000){
  if(file.exists(outfn)){
    unlink(outfn)
  }
  stopifnot(n_per_chunk %% 4 == 0)

  con <- file(description=fn, open="r")   
  data <- readLines(con, n = n_per_chunk)

  repeat {
    if (length(data) == 0)
      break

    if(length(data) %% 4 != 0){
      stop("something wrong with fastq file?")
    }

    n_lines <- length(data)
    id_lines <- seq(1, n_lines, by = 4)
    seq_lines <- seq(2, n_lines, by = 4)
    qual_lines <- seq(4, n_lines, by = 4)

    ids <- BStringSet(data[id_lines], start = 2)

    sread <- BStringSet(data[seq_lines])
    sread <- chartr("U", "T", sread)
    sread <- as(sread, "DNAStringSet")

    qual <- FastqQuality(data[qual_lines])

    fastq <- new("ShortReadQ", id = ids, sread = sread, quality = qual)

    writeFastq(fastq, outfn, mode='a')

    # last chunk was final chunk
    if (length(data) != n_per_chunk)  
      break
    # read next chunk
    data <- tryCatch(readLines(con, n = n_per_chunk), error=function(e) e)
  }

  close(con)
}

convert_rna_fastq(fq, outfq)

fq_records <- readFastq(outfq)
sread(fq_records)
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     4 TCGA

^{Created on 2022-04-20 by the [reprex package](https://reprex.tidyverse.org) (v2.0.1)}