Execution stops after skipping key processing steps

[Mangosteen24]

Thank you for developing such a valuable tool. The workflow runs successfully on the provided demo dataset; however, when I attempt to process my standard 10x single-cell v3.1 library results, the execution appears to skip the reads_processing, isoform_quantification, and apa_characterization steps, and then stops. The only output generated consists of two folders:

1. scalpeloutput/Runfiles/annotation_processing/isoform_processing/
   – Contains chromosome-level files such as .exons and .exons_unique.
2. scalpeloutput/Runfiles/annotation_processing/salmon_bulk_quantification/
   – Contains links to GTF files for each chromosome.

Could you kindly help me identify the cause of this issue? Please find my code and log file attached below for your reference.

nextflow run /sr/research/tools/SCALPEL/main.nf \
  -with-apptainer /sr/research/tools/SCALPEL.container.sif \
  --sequencing chromium \
  --outputDir scalpeloutput \
  --samplesheet /sr/research/testrun/samplesheet.csv \
  --transcriptome /sr/ref/gencode.v32.transcripts.fa \
  --gtf /sr/ref/gencode.v32.annotation.gtf \
  --barcodes /scell-data/output-data/sr/research/testrun/testdata/BXXXXXXXXXX_barcodes.txt \
  --ipdb /sr/ref/refdata-gex-GRCh38-2020-A.polyA.track --cpus 32

samplesheet.csv

BXXXXXXXXXX,/sr/research/testrun/testdata/fastq_files/BXXXXXXXXXX_1000000_S1_L001_R1_001.fastq.gz,/sr/research/testrun/testdata/fastq_files/BXXXXXXXXXX_1000000_S1_L001_R2_001.fastq.gz,/sr/research/testrun/testdata/cellranger_samples/BXXXXXXXXXX

.nextflow.log
Thank you so much!

[mschilli87]

A quick look at the log shows that reads_processing:samples_loading:read_10x (BT019329T02, BT019329T02)) runs first followed by annotation_preprocessing:salmon_transcriptome_indexing and other jobs that match your output.

It also show that the other expected jobs (reads_processing:samples_loading:bam_splitting, reads_processing:bedfile_conversion, reads_processing:reads_mapping_and_filtering, reads_processing:ip_splitting, reads_processing:ip_filtering, isoform_quantification:probability_distribution, isoform_quantification:fragment_probabilities, isoform_quantification:cells_splitting, isoform_quantification:em_algorithm, isoform_quantification:cells_merging, isoform_quantification:dge_generation, apa_characterization:differential_isoform_usage, apa_characterization:generation_filtered_bams) are started as well, but those are never mentioned again afterwards until the final all processes finished entry.

I am not sure what's going on but I would say that the least SCALPEL has an issue with error reporting here as any mistake in usage or input should raise a meaningful error rather than failing silently.

I hope @Franzx7 can help you more based on the input you already provided.

[mschilli87]

@Mangosteen24: Did you have any success troubleshooting this? If you found a solution it would be valuable to know what was the culprit so we can test for and warn about this situation in the future and/or improve our documentation. If you didn't, knowing what you tried and some more details on what happened would be useful to try to help you make it work.

[franzake]

Hi @Mangosteen24 ... and Sorry for the late reply, and thanks for the report.

If reads_processing, isoform_quantification, and apa_characterization don’t run, it usually indicates an inconsistency between the 10x reference input and the annotation (very commonly chromosome naming mismatches like 1 vs chr1, or different genome builds) that prevent the link between the annotation_processing and reads_processing tasks to happen.

The .nextflow.log alone is often not enough to diagnose this. Could you please share a very small subset of your inputs so we can reproduce?

Best,
F