The polymorphic diploid assembly of the Chinese amphioxus Branchiostoma belcheri was processed by HaploMerger. A total of 274 Mbp N gap-free alignments for trusted allele pairs (>1000-bp alignment length and >90% alignment identity) were extracted from the HaploMerger outputs. From this alignment pool, we randomly selected 10-Mbp alignments and concatenated the target and query sequences, respectively. By doing so, a small simulated diploid genome was created with a pair of 10-Mbp chromosomes, one from the target and the other from the query. A total of 25 small genomes of 10 Mbp were created. Random sampling without replacement was implemented to ensure no repeated use of any alignment from the pool. In addition, all alignments were concatenated to create a large simulated genome with a pair of 274-Mbp chromosomes.
For each chromosome from the 10-Mbp genomes, we simulated 650,000 454 shotgun reads (350 ± 70 bp), 140,000 3-kb paired-end 454 reads (3000 ± 600 bp), 40,000 8-kb paired-end 454 reads (8000 ± 1600 bp), 40,000 20-kb paired-end 454 reads (20,000 ± 4000 bp), 1,000,000 300-bp mate-pair Illumina reads (115 bp per end), and 1,000,000 500-bp mate-pair Illumina reads (115 bp per end). It should be noted that to induce more assembly errors, the length of each end of the 454 paired-end reads was set to only 104 bp. Reads were randomly sampled from the chromosomes. Sequencing errors were simulated at 1.3%–1.7% for each read. For 454 reads, ∼50% of the error rate represented indels due to homopolymers. Before use, the Illumina reads were subjected to error-correction using Quake (Kelley et al. 2010 (link)). In summary, we simulated ∼11× 454 and 23× Illumina reads for each small genome. As for the large genome, only 11× 454 reads were simulated (with the same proportions for the library size as for the small genomes). The Celera assembler version 6.1 was used to assemble the simulated data with the following specific parameters: utgErrorRate = 0.015, overlapper = mer, and unitigger = bog (Miller et al. 2008 (link)). Finally, HaploMerger was used to analyze each resulting, soft-masked assembly with the default parameters and a scoring matrix specific to the assembly.
For each chromosome from the 10-Mbp genomes, we simulated 650,000 454 shotgun reads (350 ± 70 bp), 140,000 3-kb paired-end 454 reads (3000 ± 600 bp), 40,000 8-kb paired-end 454 reads (8000 ± 1600 bp), 40,000 20-kb paired-end 454 reads (20,000 ± 4000 bp), 1,000,000 300-bp mate-pair Illumina reads (115 bp per end), and 1,000,000 500-bp mate-pair Illumina reads (115 bp per end). It should be noted that to induce more assembly errors, the length of each end of the 454 paired-end reads was set to only 104 bp. Reads were randomly sampled from the chromosomes. Sequencing errors were simulated at 1.3%–1.7% for each read. For 454 reads, ∼50% of the error rate represented indels due to homopolymers. Before use, the Illumina reads were subjected to error-correction using Quake (Kelley et al. 2010 (link)). In summary, we simulated ∼11× 454 and 23× Illumina reads for each small genome. As for the large genome, only 11× 454 reads were simulated (with the same proportions for the library size as for the small genomes). The Celera assembler version 6.1 was used to assemble the simulated data with the following specific parameters: utgErrorRate = 0.015, overlapper = mer, and unitigger = bog (Miller et al. 2008 (link)). Finally, HaploMerger was used to analyze each resulting, soft-masked assembly with the default parameters and a scoring matrix specific to the assembly.
