Closed inesmarais closed 1 year ago
JGASmits
commented
1 year ago The current seq2science pipeline requires single cell fastq files (which we use for plate based scATAC), it does not support 10x files where the files of multiple cells are concatinated in a single bam file.
Correct me if im mistaken and that is not what you are trying. So i dont think its a bug, its a missing feature.
Gr Jos
Maarten-vd-Sande
commented
1 year ago I'm not so sure, can't look into it with too much detail right now. Might be related to the fact you do not assign the sample to a technical replicate. This is your samples.tsv:
sample assembly technical_replicates
SRR11692131 hg38But perhaps something like this works?
sample assembly technical_replicates
SRR11692131 hg38 myfirsttechnicalrepEDIT: I just saw @JGASmits comment. Ignore my answer
inesmarais
commented
1 year ago @JGASmits: I used two fastq files generated by sci-ATAC-seq as input files for the pipeline to ultimately generate the bam file. The other bam file (generated by 10X) was only added as a comparison.
@Maarten-vd-Sande Does this mean that sci-ATAC-seq fastq input files do not work for the current pipeline?
Maarten-vd-Sande
commented
1 year ago Yes and no. The current implementation dates back to an ancient era where people did plate-based single-cell sequencing :sauropod: . We would then have a fastq file per cell. The current implementation expects this.
To make seq2science work with sciatac, you would have to split the fastq file into per-cell fastqs. Then probably it would work. Would I recommend this? not really! Probably easier to look for a workflow that directly supports this, or do the analysis old-school by hand. Perhaps 10x/cellranger already has a workflow specific for your needs ready to go?
JGASmits
commented
1 year ago Unless you have your data of cells as individual fastq files (with each cell having a sample file entry), the current scATAC pipeline does not work. It currently succesfully treats your BAM file as a single cell (and generates a snapfile containing this single cell).
I think snapATAC also relatively straightforward supports 10x to snapfile methods (outside of seq2science). See https://github.com/r3fang/SnapATAC/wiki/FAQs#10X_snap
Perhaps also snapATAC 2.0 has some other guidlines for integrating 10x data.
inesmarais
commented
1 year ago Thanks for both of your quick answers, it is clear to me now! Unfortunately sci-atac input is not directly supported by 10X, then we will look for an alternative!
Cheers,
Inès
Describe the bug Hi, I have used the scATAC-seq pipeline with sci-ATAC-seq input data. Although the run was indicated to be successful, the final bam file is missing cell barcodes and we think that makes the snap object contain 0 barcodes as well.
Expected behavior We expected the final bam file to have cell barcodes and the snap object to at least contain cell barcodes. The pipeline should indicate that something goes wrong if this is the case.
Screenshots Bam file compared to a bam file generated by Cellranger from 10X. The top bamfile is the bamfile where no cell barcodes can be observed.
Snapatac QC file
Seq2science log file