Demuxlet

Demuxlet is a genotype demultiplexing software that requires reference genotypes to be available for each individual in the pool. Therefore, if you don’t have reference genotypes, you may want to demultiplex with one of the softwares that do not require reference genotype data (Freemuxlet, scSplit, Souporcell or Vireo)

Data

This is the data that you will need to have prepare to run Demuxlet:

Required

• Reference SNP genotypes for each individual ($VCF)

  • Filter for common SNPs (> 5% minor allele frequency) and SNPs overlapping genes

  • Demuxlet is very sensitive to missing data in a vcf so please make sure you only have complete cases in your reference donor SNP genotype file

• Genotype field in $VCF ($FIELD)

  • This is GP by default but could also be GT others

• Barcode file ($BARCODES)

• Bam file ($BAM)

  • Aligned single cell reads

• Output directory ($DEMUXLET_OUTDIR)

Optional

• A text file with the individual ids ($INDS)

  • File containing the individual ids (separated by line) as they appear in the vcf file

  • For example, this is the individual file for our example dataset

Run Demuxlet

Popscle Pileup

First we will need to identify the number of reads from each allele at each SNP location:

With $INDS fileWithout $INDS file

The $INDS file allows demuxlet to only consider the individual in this pool

singularity exec Demuxafy.sif popscle dsc-pileup --sam $BAM --vcf $VCF --group-list $BARCODES --out $DEMUXLET_OUTDIR/pileup --sm-list $INDS

HELP! It says my file/directory doesn’t exist!

If you receive an error indicating that a file or directory doesn’t exist but you are sure that it does, this is likely an issue arising from Singularity. This is easy to fix. The issue and solution are explained in detail in the Notes About Singularity Images

If the pileup is successful, you will have these files in your $DEMUXLET_OUTDIR:

.
├── pileup.cel.gz
├── pileup.plp.gz
├── pileup.umi.gz
└── pileup.var.gz

Additional details about outputs are available below in the Demuxlet Results and Interpretation.

Popscle Demuxlet

Once you have run popscle pileup, you can demultiplex your samples:

With $INDS fileWithout $INDS file

The $INDS file allows demuxlet to only consider the individual in this pool

singularity exec Demuxafy.sif popscle demuxlet --plp $DEMUXLET_OUTDIR/pileup --vcf $VCF --field $FIELD --group-list $BARCODES --geno-error-coeff 1.0 --geno-error-offset 0.05 --out $DEMUXLET_OUTDIR/demuxlet --sm-list $INDS

HELP! It says my file/directory doesn’t exist!

If you receive an error indicating that a file or directory doesn’t exist but you are sure that it does, this is likely an issue arising from Singularity. This is easy to fix. The issue and solution are explained in detail in the Notes About Singularity Images

Note

Demuxlet by default assumes that your $VCF uses R2 to indicate the imputation score. If you have a different imputation metric (INFO is also commonly used), then you should use --r2-info to indicate the metric it should use (for example: --r2-info INFO)

If demuxlet is successful, you will have these new files in your $DEMUXLET_OUTDIR:

.
├── demuxlet.best
├── pileup.cel.gz
├── pileup.plp.gz
├── pileup.umi.gz
└── pileup.var.gz

Additional details about outputs are available below in the Demuxlet Results and Interpretation.

Demuxlet Summary

We have provided a script that will summarize the number of droplets classified as doublets, ambiguous and assigned to each donor by Demuxlet and write it to the $DEMUXLET_OUTDIR. You can run this to get a fast and easy summary of your results by providing the path to your result file:

singularity exec Demuxafy.sif bash Demuxlet_summary.sh $DEMUXLET_OUTDIR/demuxlet.best

which will return:

Classification	Assignment N
113_113	1334
349_350	1458
352_353	1607
39_39	1297
40_40	1078
41_41	1127
42_42	1419
43_43	1553
465_466	1094
596_597	1255
597_598	1517
632_633	868
633_634	960
660_661	1362
doublet	3053

or you can write it straight to a file:

singularity exec Demuxafy.sif bash Demuxlet_summary.sh $DEMUXLET_OUTDIR/demuxlet.best > $DEMUXLET_OUTDIR/demuxlet_summary.tsv

Note

To check if these numbers are consistent with the expected doublet rate in your dataset, you can use our Doublet Estimation Calculator.

Demuxlet Results and Interpretation

After running the Demuxlet steps and summarizing the results, you will have a number of files from some of the intermediary steps. These are the files that most users will find the most informative:

• demuxlet.best

  • Metrics for each droplet including the singlet, doublet or ambiguous assignment (DROPLET.TYPE), final assignment (BEST.GUESS), log likelihood of the final assignment (BEST.LLK) and other QC metrics.

    INT_ID	BARCODE	NUM.SNPS	NUM.READS	DROPLET.TYPE	BEST.GUESS	BEST.LLK	NEXT.GUESS	NEXT.LLK	DIFF.LLK.BEST.NEXT	BEST.POSTERIOR	SNG.POSTERIOR	SNG.BEST.GUESS	SNG.BEST.LLK	SNG.NEXT.GUESS	SNG.NEXT.LLK	SNG.ONLY.POSTERIOR	DBL.BEST.GUESS	DBL.BEST.LLK	DIFF.LLK.SNG.DBL
    0	AAACCTGAGATAGCAT-1	170	231	SNG	41_41,41_41,0.00	-29.42	40_40,41_41,0.50	-39.12	9.70	-33	1	41_41	-29.42	597_598	-76.24	0.00000	40_40,41_41,0.50	-39.12	9.70
    1	AAACCTGAGCAGCGTA-1	325	583	SNG	465_466,465_466,0.00	-70.61	42_42,465_466,0.50	-94.85	24.24	-74	1	465_466	-70.61	42_42	-166.61	0.00000	42_42,465_466,0.50	-94.85	24.24
    2	AAACCTGAGCGATGAC-1	147	227	SNG	113_113,113_113,0.00	-25.05	39_39,113_113,0.50	-29.85	4.80	-28	1	113_113	-25.05	349_350	-51.63	0.00000	39_39,113_113,0.50	-29.85	4.80
    3	AAACCTGAGCGTAGTG-1	180	235	SNG	349_350,349_350,0.00	-33.14	349_350,632_633,0.50	-44.78	11.64	-36	1	349_350	-33.14	632_633	-77.41	0.00000	349_350,632_633,0.50	-44.78	11.64
    4	AAACCTGAGGAGTTTA-1	248	444	SNG	632_633,632_633,0.00	-54.79	352_353,632_633,0.50	-72.23	17.43	-58	1	632_633	-54.79	633_634	-163.24	0.00000	352_353,632_633,0.50	-72.23	17.43
    …	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…

Merging Results with Other Software Results

We have provided a script that will help merge and summarize the results from multiple softwares together. See Combine Results.

Citation

If you used the Demuxafy platform for analysis, please reference our preprint as well as Demuxlet.