BS-seq about SNP_split

[zhangaicen]

Dear Felix， recently I tried to use the SNP_split to split BS-seq reads from a rice hybrid, in the user guide I see that for SNP_split it adatped the bam from bismark, and before using SNP_split, the SNP position should be masked with "N", but when using the bismark, it will firstly convert "C" to "T", "G" to "A" in the BS-seq genome preparation and mapping step, so is the SNP positions masked by "N" are excluded for methylation level caculation because of no conversion?

Waiting for your reply, thanks.

[FelixKrueger]

Hi @zhangaicen

Yes, the workflow for bisulfite sequencing would be:

1. Create an N-masked genome reference (for the mouse genome this would work out-of-the-box with SNPsplit_genome_preparation. but for a (phased?) rice hybrid genome you would have to come up with a solution for this yourself).
2. Index the N-masked genome
3. Align with Bismark to the N-masked genome
4. Deduplicate
5. Spit allele-specifically using SNPsplit#

Indeed, positions aligning to Ns in the genome will not receive a methylation call (only positions with a C in the genome are called). The behaviour in SNPsplit is a little different though: if the read aligns to the strand containing an N-masked C, and the SNP was a C to T transition, the position is not used for allele-assignment (as the information could either be a SNP or a methylation state). If the read aligns to the strand opposite the C, the position may be used for allele-assignment.

Does this help?

[zhangaicen]

Yes，I‘ve finished the step1-step4, so considering the whole genome, the methytion level of SNP site may not have to much impact, so I think this can be ignored， all right?

[FelixKrueger]

I wouldn't think that the SNPs matter very much for the global picture, and at least half of reads over C to T SNPs can still be used for allele-assignments. Good luck!