[Issue] How to deal with TAPS data using biscuit?

[YuanfengZhang]

In the paper, it's said that:

Besides bisulfite conversion, BISCUIT offers integrated computational solutions for analyzing other epigenomic profiling data derived from cytosine conversion principles, such as NOMe-seq, TAB-seq, ACE-seq, TAPS and hairpin-based techniques like five- and six-letter sequencing.

However I did not find any available params for TAPS in biscuit align as well as other sub-cmdln documentations. Does it mean that biscuit can automatically infer the base conversion strategy?

[jamorrison]

For alignments, the specific method used to perform the cytosine conversion doesn't matter for BISCUIT. It will align any cytosine converted (C→T) dataset.

For TAPS specifically, I have not processed TAPS data myself, so I would suggest starting with the default algorithm and scoring options. If you are seeing poor alignment rates, these may need to be adjusted, though I would double check data quality, proper ordering of FASTQs, etc. before adjusting these options.

[YuanfengZhang]

Thx, I will give it a try.

the specific method used to perform the cytosine conversion doesn't matter for BISCUIT

Do you mean the mC→T / unmodC→T doesn't matter?

[jamorrison]

I was specifically referring to whether the conversion is chemical (e.g., sodium bisulfite) or enzymatic (e.g., EM-seq). BISCUIT will align both mC→T and unmodC→T to a reference genome as biscuit align will allow T's in reads to align to C's in the reference regardless of whether the conversion is from a methylated/modified cytosine or an unmodified cytosine.

Other subcommands in BISCUIT, though, work under the assumption that converted cytosines are the result of unmethylated cytosines being converted (C→T). Therefore, I would caution against using the other subcommands without first understanding how this assumption may impact analyzing mC→T converted data.

[YuanfengZhang]

Update: biscuit works well for TAPS data. For align, nothing need to be changed. For count, just filp the methylated count and unmethylated count in the epibed.
It should be something like that:

biscuit vcf2bed ${biscuit_vcf} |\
awk 'BEGIN{FS=OFS="\t"} {print $1, $2, $3, $4, $5, $6, $7, $10, $9}' |\
pigz --best > ${biscuit_epibed}

[jamorrison]

I'm glad to hear BISCUIT does well with TAPS data!

Can I confirm which file format you're talking about? biscuit vcf2bed outputs the BISCUIT BED file (chr, start, end, beta value, coverage), while biscuit epiread generates an epiBED file (chr, start, end, read name, read number, bisulfite strand, CpG methylated/unmethylated status, GpC methylated/unmethylated status, SNP/indel status).

[YuanfengZhang]

I'm glad to hear BISCUIT does well with TAPS data!

Can I confirm which file format you're talking about? biscuit vcf2bed outputs the BISCUIT BED file (chr, start, end, beta value, coverage), while biscuit epiread generates an epiBED file (chr, start, end, read name, read number, bisulfite strand, CpG methylated/unmethylated status, GpC methylated/unmethylated status, SNP/indel status).

vcf2bed