Issues with scsplit count

[yilevine]

Hello!

I was trying to use scSplit on snRNA-seq data with 4 mixed samples. But I always got errors with the code below.

scSplit count -c $VCF/common_snvs_hg38_scsplit -v $SCSPLIT_OUTDIR/freebayes_region_var_qual30.recode.vcf -i $SCSPLIT_OUTDIR/filtered_bam_dedup_sorted_named.bam -b $BARCODES/barcodes_merged_IDCM_20pc.tsv -r $SCSPLIT_OUTDIR/ref_filtered.csv -a $SCSPLIT_OUTDIR/alt_filtered.csv -o $SCSPLIT_OUTDIR

I suspected the problem might be that the vcf file is not one sample but rather a mixed sample mentioned in #6 and #19. So I checked my vcf file which was generated by freebayes freebayes-parallel targets.regions 27 -f $FASTA/genome.fa $SCSPLIT_OUTDIR/filtered_bam_dedup_sorted_named.bam>freebayes_region_var.vcf

Meanwhile, I also checked my bam file.

Therefore, I believe the problem is my vcf file generated by freebayes.

But I didn't understand why I got a mixed vcf file Do you have any idea on how to fix it?

Thanks! Yile

[jon-xu]

Hi Yile,

I guess it's related with your RG tag in the BAM file? Check this out and you might want to contact freebayes for more details: https://github.com/freebayes/freebayes/blob/master/README.md

Jon

[yilevine]

Hi Jon,

Thanks for your reply. I will check it out.

Best, Yile

[yilevine]

Hi Jon,

I have solved the RG tag problem. Below are my new bam file and vcf file.

So I moved forward. However, I have another problem when running scSplit count. singularity exec --bind $SINGULARITY/heart $SINGULARITY/Demuxafy.sif scSplit count -c $VCF/common_snvs_hg38_scsplit -v $SCSPLIT_OUTDIR/freebayes_region_var_single_qual30.recode.vcf -i $SCSPLIT_OUTDIR/filtered_bam_dedup_sorted_named_RG.bam -b $BARCODES/barcodes_merged_IDCM_20pc.tsv -r $SCSPLIT_OUTDIR/ref_filtered.csv -a $SCSPLIT_OUTDIR/alt_filtered.csv -o $SCSPLIT_OUTDIR I allocated 150 GB of memory and 30 hours. But I didn't get the results. Below is the scSplit.log

I don't know what is the problem with my input files. Or should I let it run longer?

Thanks. Yile

[jon-xu]

Sorry Yile for the late reply. I’m not sure about demuxafy. But your number of barcodes and positions look both quite big, which might lead to the performance issue.

Please: 1) check the freebase parameters and filter criteria mentioned in our documentation - most cases we see around 100,000 informative SNVs. 2) use the most updated version of scSplit on GitHub directly, as we solved some performance issue in the newest release.

Hope it helps, but it’s true that we didn’t test that many cells in our development.

[yilevine]

Hi Jon,

Thanks very much for your reply. The problem was indeed the number of cells. After filtration, I successfully ran,

Best, Yile