Hi all, there was this issue posted some time back about using single-end data: https://github.com/nf-core/scrnaseq/issues/394 I just got some single-end data, but the pipeline is unhappy even if I leave a blank spot for "fastq_2". Is there any way I can use this pipeline for single-end?

A couple of related threads • Your own one: https://nfcore.slack.com/archives/CHN5BV5DW/p1755699080015829 • https://nfcore.slack.com/archives/CHN5BV5DW/p1722922415436879 • https://nfcore.slack.com/archives/CHN5BV5DW/p1701705890456409 So usually if people ask to this channel how to handle single-end reads, there is usually a misconception involved. If you think in your case you actually only have single-end reads, please provide more details (sequencing technology, link to dataset if it is public)

Yes, you're right. [osu7224@ascend-login01 Eric_scRNAseq_Practice]$ gunzip -c HRR232020.fastq.gz | head @A0017420HFV22DSXX4213088564899 CB:AGGCATTCAAGATTGA-1 UB:GGTAAAAACTAG CCTCAGTAATCAGAAAAGCCGGGCTCGGCCGCCCCTTGCTTGCAGCCGGGCACTACAGGACCGGCTTGCTCACGGTGCTGTGCCAGGGCGCCCCCTGCTGGCGACTAGGGCAACTGCAGGGCTCTCTTGCTTAGACTGGTGGCCAGCGCCC + FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF,FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF @A0017420HFV22DSXX42147222271877 CB:GCTTGGGTCATTATCC-1 UB:ACACATGTTTTC CTACAGGACCGGCTTGCTCACGGTGCTGTGCCAGGGCGCCCCCTGCTGGCGACTAGGGCAACTGCAGGGCTCTCTTGCTTAGACTGGTGGCCAGCGCCCCCTGCTGGCGCCGGGGCACTGCAGGGCCCTCTTGCTTACTGTATAGTGGTGG + FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF,FFFFFFFFF @A0017420HFV22DSXX41249351316939 CB:CATACCCCAACCTAAC-1 UB:TTATGGGATGTG CCCCTGCTGGCGCCGGGGCCCTGCAGGGCCCTCTAGCTTACTGGATAGTGGTGGCACGCCGCCTGCTGGCAGCTAGGGACCTTGCAGGGTCCACTTGCTCAAGGGGTAGTGGCAGCACGCCCCCCTGCTGCCAGCTGGGGACACTCCCGGG So, clearly I need to separate out the CB and UB from the rest of the FASTQ. So I need to put the CB and UB in an R1 file and the actual transcript as R2, right?

Usually these FastQ files are constructed following a sequencing protocol like 10Xv3 or others. The read geometries for the different protocols are fixed, known, and encoded in the pipeline. The FASTQ you provide seems to match none of the standard protocols, but the barcodes/UMIs look 10X-like. It seems a bit like the FASTQ was created by extracting reads from an already-processed BAM file.