Fig. 1

DISMISS utilises strand biases generated during MeDIP-Seq library preparation to detect strand-specific DNA methylation. a Outline of MeDIP-Seq library preparation. (I) Genomic 5mC can be present in three possible scenarios: (left) with 5mC found on both DNA strands, (middle) with 5mC found only on the plus DNA strand (black coloured strand) and (right) with 5mC found only on the minus DNA strand (grey coloured strand). (II) DNA is fragmented, end-repaired and the forked adapters ligated directionally to the DNA fragments – solid black (5′ adapter) and dashed black (3′ adapter); (III) DNA is denatured. (A-IV) MeDIP enrichment (grey spheres) selects single-stranded 5mC containing fragments; a sampling bias arises due to the selection of strands originating from DNA fragments containing 5mC on only a single strand (middle and right) – note the black adaptor is retained, identifying the strand origin of the selected fragment. (V) PCR library enrichment is performed (5mC lost during PCR amplification and replaced with C). (VI) Black arrows show sequencing of first mate reads (containing the MeDIP selected 5mC) occurring from the black adaptor. DNA fragments from all three scenarios are sequenced from both ends, with those originating from black and grey adaptors deposited in a single pair of first and second FASTQ files, respectively. Reads from the minus strand will tend to align to the genome sequence in reverse-compliment mode. b C to G ratio bias detected in a MeDIP-Seq FASTQ file. Base count per cycle plot for the first FASTQ file. The x-axis shows the base position in the read, while the y-axis shows the count of each nucleotide at that position. The A to T ratios do not show a bias, as is expected for a random library not selecting for either of these nucleobases. However, the C to G ratio demonstrates a clear bias due to the selection of 5mC containing fragments (in step A-IV)