ly la ti o n c al l a cc u ra cy , % 100 - bp 76 - bp , strict parameters 76 - bp , loose parameters

Supplemental Figure 2 Mapping accuracy as a function of the percentage of uniquely mapped reads Supplemental Figure 3 Methylation call accuracy Supplemental Figure 4 Methylation level accuracy Supplemental Table 1 The results of mapping 10,633,033 real 101-bp long single-end reads from human genome, SRR306435, mate 1 Supplemental Table 2 The results of mapping 1M real 101-bp long paired-end reads from human genome, SRR306435 Supplemental Table 3 The results of mapping 1M real 76-bp long single-end reads from human genome, SRR306421 Supplemental Table 4 Mapping accuracy test: 1M synthetic 100bp single-end reads generated from human genome GRCh38 with 5% of sequencing errors and 2% of SNPs Supplemental Table 5 Mapping accuracy test: 500,016 synthetic 100bp paired-end reads generated from human genome GRCh38 with 2% of sequencing errors, 1% of SNPs, and with a single indel introduced to 5% of reads, and adapter sequence up to 15bp long replaced 3'-end of 10% of the reads Supplemental Table 6 Mapping accuracy test: 1M synthetic 76bp single-end reads with 3% of sequencing errors, 1% of SNPs, 10% of all reads having indels, and 30% of reads having an adapter sequence up to 15bp at the 3'-end of the reads Supplemental Table 7 Methylation call and level accuracy test: 20M synthetic 100bp single-end reads from chromosome 21 of the human genome GRCh38 with 5% of sequencing errors and 2% of SNPs Supplemental Table 8 Methylation call and level accuracy test: 20M synthetic 76bp single-end reads from chromosome 21 of the human genome GRCh38 with 5% of sequencing errors and 2% of SNPs, indels of length up to 10bp introduced to 10% of reads, adapter sequences up to 15bp replaced 3'-end of 10% of reads Supplemental Table 9 Comparison of the performance of BRAT-nova on real data sets using local alignment versus full-length alignment Supplemental Table 10 Feature comparison of BRAT-nova, Bismark, BS-Seeker2 and BRAT-BW Supplemental Notes BRAT-nova design and benchmarking