Libraries for whole genome bisulfite sequencing have been created using the Epignome bisulfite kit from Epicentre. I have been told that there can be a bias in fragment selection using this kit based on CG content and other larger-features. To investigate the results of our libraries, sequencing was performed on a MiSeq to determine how it looks compared to a publicaly available wgbs dataset for brachypodium (SRR628921):

 

MiSeq reads (SE 50) were grouped into ‘allmiseq’ reads to look for overall sequence variation across the dataset. Reads were trimmed and aligned to the Bd21Control reference genome from brachypodium.org. Resulting alignments were used to determine overall coverage across the genome (bedtools genomecov) and mapped to the nearest annotated gene and repeat in the genome.

coverage_genescoverage_repeats

 

These quick plots display the average count of reads within either genes or repeats (limited by dashed lines) as well as the surrounding 1kb of sequence. The ‘allmiseq’ are the epignome data and the ‘gaut’ lines are from SRA. All reads were mapped using the same pipeline. The “_bga” is just an analysis test to include zero-coverage regions in the genomecov bed file that is created.

If anything, there is less uniformity in the gaut data compared to the epignome data, however this is in part due to the difference in overall read number.

As far as visualizing the methylation data, it appears the epignome kit performed as expected:

igv1

 

 

Clear enrichment for heterochromatin in centromeric (gene-poor) regions is clearly visable. This pattern is observed with between 1.4 and 2.8M reads per sample.

When looking at smaller regions, diversity appears clear and there are minimal clonal reads:

igv3

 

When this is compared to the (higher depth) gaut bisulfite data (bright blue tracks), we see that although the miseq has lower coverage (as expected), we are basically hitting the same genomic regions with our alignments.

 

From this, it would seem that the Epignome kit is working as described. Any biases may be observed when comparing GC content, however that is still to come. Also, the addition of increased depth may clarify things.

Share →

Leave a Reply

Your email address will not be published. Required fields are marked *