Parse Bioscience WT

[james-obrien87]

Hi,

I’m interesting in using FLAMES to process an ONT sequencing run on a Parse Biosciences WT library.

Read structure is:
P5–UDI–R1–cDNA Insert–BC1–L1–BC2–L2–BC3–PolyN–R2–UDI–P7

P5-[8bp]-R1-cDNA Insert-[8bp]-[12bp]-[8bp]-[12bp]-[8bp]-[10bp]-R2-[8bp]-P7

The 3 barcodes each 8bp longs are joined with 2 linker sequences each 12bp long.

Is there a way to define a custom barcode structure to use with what demuliplexing etc… ?

Unlike 10x the Read 1 contains the cDNA insert whilst R2 contains B1,2&3.

I would be grateful if anyone can shed any light on this.

[ChangqingW]

Hi James,

I have been working on a fork of flexiplex to handle multiple barcodes and testing it with LR-split-seq data. I am planning to incorporate it into FLAMES but it will take me a month or two to get all the build checks passed, but you could demultiplex it manually with the modiefied flexiplex and use it as input to FLAMES (specifying pipeline_parameters.do_barcode_demultiplex = FALSE)

[ChangqingW]

For LR-split-seq you could do:

./flexiplex -p 8 -u '??????????' \
    -b '????????,list:barcodes/bc1.csv,buffer:2' -x 'GTGGCCGATGTTTCGCATCGGCGTACGACT' \
    -b '????????,list:barcodes/bc2.csv,buffer:2:' -x 'ATCCACGTGCTTGAGACTGTGG' \
    -b '????????,list:barcodes/bc3.csv,buffer:2' -x 'TTTTTTTT' \
    -e 2 -f 8 sequencing_data/LR-Split-seq/SRR13948564.fastq \
    > LR-split-seq-demultiplexed.fastq"

For the pattern you provided you could try:

./flexiplex -p 8 \
    -b '????????,list:barcodes/bc1.csv,buffer:2' -x 'L1 sequence' \
    -b '????????,list:barcodes/bc2.csv,buffer:2:' -x 'L2 sequence' \
    -b '????????,list:barcodes/bc3.csv,buffer:2' -x 'TTTTTTTT' \
    -e 2 -f 8 in.fastq > out.fq

-p is for number of threads, -e for edit-distances each barcode segment could have (or you could specify them individually with -b '????????,list:barcodes/bc2.csv,buffer:2,max_ed:2'), and -f is for the max edit-distances for the L1 + L2 + polyT pattern alignment.

Does this protocol not have an UMI sequence? I don't think our gene quantification works without UMIs.

[james-obrien87]

Thanks I will try flexiplex to demultiplex the reads.

Whats the best name to use for the output of flexiplex to work in the FLAMES pipeline (using pipeline_parameters.do_barcode_demultiplex = FALSE) ?

Yes there is a UMI, it's what the PolyN tail is in the structure that I posted.

[james-obrien87]

@ChangqingW I used your fork of flexiplex, and ran the following command:

Flexiplex \
    -p 16 \
    -b '????????,name:BC1,list:barcodes/bc1.csv,buffer:2' -x 'GTGGCCGATGTTTCGCATCGGCGTACGACT' \
    -b '????????,name:BC2,list:barcodes/bc2.csv,buffer:2' -x 'ATCCACGTGCTTGAGACTGTGG' \
    -b '????????,name:BC3,list:barcodes/bc3.csv,buffer:2' \
    -u '??????????' \
    -e 3 -f 8 \
    in.fastq > out.fastq

The in.fastq has a total of 10,512,838 sequences, whilst the demultiplex out.fastq only has 69,937 which seems much lower than expected.

Theres approx 5k cells which would mean that theres only 13.9 reads per cell.

Not sure if I'm doing something wrong or this is the underlying data.

The barcode files come straight from the parse pipeline.