Evaluation of RAFT-hifiasm using simulated data
| PacBio HiFi | ONT Simplex | ONT Duplex | ||||
|---|---|---|---|---|---|---|
| 30× | 30× | 50× | 30× | 50× | ||
| Assembly gaps due to contained read deletion | Standard string graph | 5 | 51 | 11 | 30 | 10 |
| hifiasm | 3 | 40 | 6 | 17 | 4 | |
| RAFT-hifiasm | 0 | 1 | 0 | 2 | 1 | |
| Fraction of bases used in string graph | Standard string graph | 44.54% | 23.87% | 17.94% | 30.91% | 22.93% |
| hifiasm | 44.55% | 24.09% | 18.02% | 30.92% | 22.94% | |
| RAFT-hifiasm | 73.27% | 61.01% | 57.22% | 65.88% | 62.04% | |
[i] We show the count of assembly gaps in the string graphs constructed by three methods, and the fraction of bases used in the graphs. The standard string graph is based on the original string graph formulation (Myers 2005) that ignores contained reads. The row labeled hifiasm is the string graph, which contains contained reads that were rescued by a heuristic implemented in hifiasm. The row RAFT-hifiasm considers the string graph constructed after fragmenting reads using RAFT and then assembling the fragmented reads using hifiasm.