The nf execution directory does not match the pwd. Rather, the wd of the main.nf is used.

Because the seqtools.R script takes all fastq as arguments, it does not parallelize natively with nextflow.
Use parallel in R

Zig-zag curves...

Go recursively in input folder (--readsdir) and include all fastq files found in the report. Simple change in the search pattern to **_R{1,2}_001.fastq.gz should be enough.

The multiqc report file is published OK, this message is due to FileAlreadyExistsException. When the pipeline is run for a second time (and multiqc report is already there), no warning. Check script if there is somehow duplication of the published file.

process multiqc fails with big numbers of reads or bases. Presumably the formatter is causing this (java.text.DecimalFormat)

For big fastq files, the N50 calculation gets an integer overflow in 'cumsum'

This here:
Reduce( mergeFastqq, mclapply(params$fqfiles, fastqq, k = 3) )

causes this error with Nanopore reads:

Quitting from lines 30-91 (fastq-stats-report.Rmd)
  Error in cbind(lhs@seqLenCount, rhs@seqLenCount) :
    number of rows of matrices must match (see arg 2)
  Calls: <Anonymous> ... withVisible -> eval -> eval -> Reduce -> f -> f -> cbind