3.5. Testing optimized in silico method using simulated ancient DNA reads
The quality of the genome assembly of T. flavidus generated using only short paired-end libraries was unsatisfactory (N50: 0.8 Kb, complete BUSCOs: 148); (Table 4). When conserved in silicomate-pair libraries were generated using two genus references, compared to the original in silico mate-pair libraries using one reference, the NGA50 increased (NGA50: aDNA-rub-bim**: 438.4 Kb vs. aDNA-rub*:354.3 Kb), whereas misassemblies decreased significantly (misassemblies: aDNA-rub-bim**: 985 vs. aDNA-rub*: 1,661) and comparable numbers of complete genomes (complete BUSCOs: aDNA-rub-bim**: 2,156 vs. aDNA-rub*: BUSCOs: 2,205).
Genome assembly using the RaGOO pipeline showed higher contiguity (NGA50: aDNA-rub@: 727.7 Kb vs. aDNA-rub-bim**: 438.4 Kb, @: assemblies using RaGOO method) and higher gene completeness (complete BUSCOs: aDNA-rub@: 2,203 vs. aDNA-rub-bim**: 2,156), but with many more errors (misassemblies: aDNA-rub@: 1,829 vs. aDNA-rub-bim**:985), compared to using conserved in silicomate-pair libraries generated using two genera references. Synteny between assemblies and “real” genome (the best assembly of T. flavidus ) using the optimized in silico method was better than that using the RaGOO method (Figs. S2-S3).