Thank you for your tool, but I found that the long-read single-cell RNA-seq analysis tutorial does not seem to match the FLAMES output.

[haili2717096-lab]

I found that files such as ./data/background/gene_count.csv do not exist in the FLAMES output, and the long-read single-cell RNA-seq analysis tutorial does not seem to match the FLAMES output.

[ChangqingW]

Hi @haili2717096-lab,
Thank you for your interest. The package is being updated frequently recently so some documentation may be outdated. I was wondering which specific tutorial you are referring to so we could update it.
If you were referring to @Sefi196 's one, I would assume the data should be downloaded from his repository at https://github.com/Sefi196/FLAMESv2_LR_sc_tutorial/tree/main/data/background. It does seem that the wget command in Section 1.2 is malformed. You could either manually download it or use git clone git@github.com:Sefi196/FLAMESv2_LR_sc_tutorial.git && mv FLAMESv2_LR_sc_tutorial/data . && rm -r FLAMESv2_LR_sc_tutorial

[Sefi196]

Hi @haili2717096-lab

The data folder can be downloaded here

wget https://github.com/Sefi196/FLAMESv2_LR_sc_tutorial/archive/refs/heads/main.zip
unzip main.zip "FLAMESv2_LR_sc_tutorial-main/data/*"

I will update the repo accordingly.

If your after background/gene_count.csv you will need to run FLAMESv2 on a set of background barcodes to generate these counts. FLAMES does not run this analysis yet by default. We do plan to add it in as feature at some stage. In the meantime a description of how do so is found here in the tutorial https://sefi196.github.io/FLAMESv2_LR_sc_tutorial/appendix.html#calculating-the-ambient-rna-profile

let me know if you have any other questions

If your knee plot looks good you can skip this step as removal of empty droplets and ambient RNA is likely not required.

[haili2717096-lab]

In the Function Reference Manual, I found that MultiSampleSCPipeline is deprecated and seems to have been replaced by sc_long_multisample_pipeline. However, when I use sc_long_multisample_pipeline, I encounter the error object 'expect_cell_numbers' not found.

According to the R help documentation, expect_cell_numbers should only be required when barcodes_file is not provided. But in my case, even though I supplied barcodes_file, the function still requires expect_cell_numbers.

Will this requirement affect my results?I found that files such as ./data/background/gene_count.csv do not exist in the FLAMES output

[haili2717096-lab]

I found that even though I provided the barcodes_file, it still performs "Searching for barcodes".

sce_list <- FLAMES::sc_long_multisample_pipeline( + annotation = annotation, + fastqs = fastqs, + outdir = outdir, + genome_fa = genome_fa, + minimap2 = minimap2_bin, + k8 = k8_bin, + barcodes_file = barcodes_file, + expect_cell_numbers = 10000 + ) No config file provided, creating a default config in /home/user3/scRNAseq/data/Nanopore/combind Writing configuration parameters to: /home/user3/scRNAseq/data/Nanopore/combind/config_file_176335.json FLEXIPLEX 0.96.2 Setting max barcode edit distance to 2 Setting max flanking sequence edit distance to 8 Setting read IDs to be replaced Setting number of threads to 8 Search pattern: primer: CTACACGACGCTCTTCCGATCT BC: NNNNNNNNNNNNNNNN UMI: NNNNNNNNNNNN polyT: TTTTTTTTT Number of known barcodes: 1 Processing file: /home/user3/scRNAseq/data/row_data/nanopore/DT-pituitary/DT-small/20231204-NPL2300768-P4-PAS56794-sup.pass.fastq.gz Searching for barcodes... 0.1 million reads processed.. 0.2 million reads processed.. 0.3 million reads processed..

[ChangqingW]

In the Function Reference Manual, I found that MultiSampleSCPipeline is deprecated and seems to have been replaced by sc_long_multisample_pipeline. However, when I use sc_long_multisample_pipeline, I encounter the error object 'expect_cell_numbers' not found.

According to the R help documentation, expect_cell_numbers should only be required when barcodes_file is not provided. But in my case, even though I supplied barcodes_file, the function still requires expect_cell_numbers.

Will this requirement affect my results?I found that files such as ./data/background/gene_count.csv do not exist in the FLAMES output

sc_long_multisample_pipeline is deprecated and replaced by MultiSampleSCPipeline..

./data/background/gene_count.csv from Sefi's tutorial is expected to be downloaded whereas FLAMES would output a gene_counts.csv or multiple [smaple_name]_gene_counts.csv (with multi sample pipeline) in the output folder if do_gene_quantification is set to true (which is the default).

[ChangqingW]

I found that even though I provided the barcodes_file, it still performs "Searching for barcodes".

sce_list <- FLAMES::sc_long_multisample_pipeline( + annotation = annotation, + fastqs = fastqs, + outdir = outdir, + genome_fa = genome_fa, + minimap2 = minimap2_bin, + k8 = k8_bin, + barcodes_file = barcodes_file, + expect_cell_numbers = 10000 + ) No config file provided, creating a default config in /home/user3/scRNAseq/data/Nanopore/combind Writing configuration parameters to: /home/user3/scRNAseq/data/Nanopore/combind/config_file_176335.json FLEXIPLEX 0.96.2 Setting max barcode edit distance to 2 Setting max flanking sequence edit distance to 8 Setting read IDs to be replaced Setting number of threads to 8 Search pattern: primer: CTACACGACGCTCTTCCGATCT BC: NNNNNNNNNNNNNNNN UMI: NNNNNNNNNNNN polyT: TTTTTTTTT Number of known barcodes: 1 Processing file: /home/user3/scRNAseq/data/row_data/nanopore/DT-pituitary/DT-small/20231204-NPL2300768-P4-PAS56794-sup.pass.fastq.gz Searching for barcodes... 0.1 million reads processed.. 0.2 million reads processed.. 0.3 million reads processed..

That is the output from flexiplex, meaning searching for the barcodes from the barcode list in the reads, i.e. demultiplexing reads with the given barcode allow-list. I am more concerned by the Number of known barcodes: 1 message, it looks like the barcode file path you provided was not accessible to flexiplex and it treated as a barcode string itself.

[ChangqingW]

pipeline_parameters.demultiplexer = "flexiplex" is required when creating the config with create_config (or you could manually update the config file) to use flexiplex instead of BLAZE (BLAZE would require the expcted cell number and infer putative barcodes instead of using the barcode list) as per the latest development version from github. This is not yet documented, I'll update the reference for MultiSampleSCPipeline shortly. Thanks for picking this up and sorry for the confusion.

[haili2717096-lab]

Thank you very much for your answer. However, I have encountered another issue when trying to install the latest development version from GitHub:

package ‘Seqinfo’ is not available for Bioconductor version '3.21'
A version of this package for your R version might be available elsewhere.

I noticed that using the Bioconductor development channel might work, but it would update a large number of already installed packages, which could cause compatibility issues. Do you have any good solutions for this?

[ChangqingW]

You could keep using the release version. I have not added the pipeline_parameters.demultiplexer option to it so you can stick with the reference, if I recall correctly.

[haili2717096-lab]

Thank you for your reply. However, I am encountering this issue in the released version. I can confirm that the data and reference genome are correct — I have already validated them using the Python implementation of FLAMES — and I need a solution for the R version.

[M::worker_pipeline::6584.84731.81] mapped 638043 sequences
[M::worker_pipeline::6611.35731.77] mapped 638621 sequences
[M::worker_pipeline::6628.681*31.68] mapped 415399 sequences
[M::main] Version: 2.30-r1287
[M::main] CMD: /home/user3/miniconda3/envs/Minimap2/bin/minimap2 --eqx -N 100 -ax map-ont -y -t 80 /home/user3/scRNAseq/data/Nanopore/combind/transcript_assembly.fa /home/user3/scRNAseq/data/Nanopore/combind/DSE_matched_reads_dedup.fastq
[M::main] Real time: 6629.722 sec; CPU: 210019.095 sec; Peak RSS: 12.038 GB
Sorting by CB for oarfish quantifaction
Sorting BAM file with 80 threads...
[bam_sort_core] merging from 8 files and 80 in-memory blocks...
Running minimap2 with args: --eqx -N 100 -ax map-ont -y -t 80

Generating transcript count matrix

2025-08-28T23:20:09.544843Z INFO oarfish: setting user-provided filter parameters.
2025-08-28T23:20:09.761978Z INFO oarfish::alignment_parser: read header from BAM file /home/user3/scRNAseq/data/Nanopore/combind/DT.small_realign2transcript.bam, contains 295,302 reference sequences.
2025-08-28T23:20:09.762002Z INFO oarfish::alignment_parser: saw minimap2 as a program in the header; proceeding.
2025-08-28T23:20:09.762005Z INFO oarfish: calculating seqcol digest
2025-08-28T23:20:09.845052Z INFO oarfish: done calculating seqcol digest
2025-08-28T23:20:09.866604Z INFO oarfish: parsed reference information for 295,302 transcripts.
Error in DataFrame(x = seq_len(ncol(a1)), row.names = nms) :
missing values in 'row.names'
8.
stop("missing values in 'row.names'")
7.
DataFrame(x = seq_len(ncol(a1)), row.names = nms)
6.
SummarizedExperiment(...)
5.
SingleCellExperiment::SingleCellExperiment(assays = list(counts = mtx))
4.
parse_oarfish_sc_output(oarfish_out, annotation, outdir)
3.
quantify_transcript_oarfish(annotation, outdir, config, pipeline,
...)
2.
quantify_transcript(annotation = annotation, outdir = outdir,
config = config, pipeline = "sc_multi_sample", samples = names(fastqs))
1.
FLAMES::sc_long_multisample_pipeline(annotation = annotation,
fastqs = fastqs, outdir = outdir, genome_fa = genome_fa,
minimap2 = minimap2_bin, k8 = k8_bin, barcodes_file = barcodes_file,
expect_cell_numbers = 10000)

[ChangqingW]

Created a separate issue for this in #63 as this is a software bug and less relevant to Sefi's tutorial.

Thank you for your reply. However, I am encountering this issue in the released version. I can confirm that the data and reference genome are correct — I have already validated them using the Python implementation of FLAMES — and I need a solution for the R version.

[M::worker_pipeline::6584.847_31.81] mapped 638043 sequences [M::worker_pipeline::6611.357_31.77] mapped 638621 sequences [M::worker_pipeline::6628.681*31.68] mapped 415399 sequences [M::main] Version: 2.30-r1287 [M::main] CMD: /home/user3/miniconda3/envs/Minimap2/bin/minimap2 --eqx -N 100 -ax map-ont -y -t 80 /home/user3/scRNAseq/data/Nanopore/combind/transcript_assembly.fa /home/user3/scRNAseq/data/Nanopore/combind/DSE_matched_reads_dedup.fastq [M::main] Real time: 6629.722 sec; CPU: 210019.095 sec; Peak RSS: 12.038 GB Sorting by CB for oarfish quantifaction Sorting BAM file with 80 threads... [bam_sort_core] merging from 8 files and 80 in-memory blocks... Running minimap2 with args: --eqx -N 100 -ax map-ont -y -t 80

Generating transcript count matrix

2025-08-28T23:20:09.544843Z INFO oarfish: setting user-provided filter parameters. 2025-08-28T23:20:09.761978Z INFO oarfish::alignment_parser: read header from BAM file /home/user3/scRNAseq/data/Nanopore/combind/DT.small_realign2transcript.bam, contains 295,302 reference sequences. 2025-08-28T23:20:09.762002Z INFO oarfish::alignment_parser: saw minimap2 as a program in the header; proceeding. 2025-08-28T23:20:09.762005Z INFO oarfish: calculating seqcol digest 2025-08-28T23:20:09.845052Z INFO oarfish: done calculating seqcol digest 2025-08-28T23:20:09.866604Z INFO oarfish: parsed reference information for 295,302 transcripts. Error in DataFrame(x = seq_len(ncol(a1)), row.names = nms) : missing values in 'row.names' 8. stop("missing values in 'row.names'") 7. DataFrame(x = seq_len(ncol(a1)), row.names = nms) 6. SummarizedExperiment(...) 5. SingleCellExperiment::SingleCellExperiment(assays = list(counts = mtx)) 4. parse_oarfish_sc_output(oarfish_out, annotation, outdir) 3. quantify_transcript_oarfish(annotation, outdir, config, pipeline, ...) 2. quantify_transcript(annotation = annotation, outdir = outdir, config = config, pipeline = "sc_multi_sample", samples = names(fastqs)) 1. FLAMES::sc_long_multisample_pipeline(annotation = annotation, fastqs = fastqs, outdir = outdir, genome_fa = genome_fa, minimap2 = minimap2_bin, k8 = k8_bin, barcodes_file = barcodes_file, expect_cell_numbers = 10000)