Table 4.
Summary of de novo assembly results on light chains of three antibody data sets using the de novo peptide sequencing tools SMSNet, PointNovo, Casanovo, and the de Bruijn assembler ALPS (k = 7). We used the Top 20 contigs to compare the length, coverage and accuracy of mapped contigs. Mapped contigs must be aligned to the reference protein sequence. The longest contig describes the maximum length of all generated contigs. Sequence coverage was calculated as the percentage of amino acids of the complete protein sequence that was covered by at least one contig. Accuracy was calculated as the percentage of all protein sequence calls that were labeled correctly
| IgG1 LC (216 AA) | WIgG1 LC (219 AA) | Herceptin LC (214 AA) | |
|---|---|---|---|
| SMSNet | |||
| Mapped contigs | 10 | 5 | 8 |
| Longest contig | 51 (23.61%) | 61 (27.86%) | 67 (31.30%) |
| Sequence coverage | 196 (90.74%) | 200 (91.32%) | 208 (97.20%) |
| Sequence accuracy | 171 (87.24%) | 190 (95.00%) | 183 (87.98%) |
| PointNovo | |||
| Mapped contigs | 7 | 3 | 6 |
| Longest contig | 51 (23.61%) | 108 (49.32%) | 75 (35.05%) |
| Sequence coverage | 205 (94.91%) | 204 (93.15%) | 212 (99.07%) |
| Sequence accuracy | 187 (91.22%) | 191 (93.63%) | 190 (89.62%) |
| Casanovo | |||
| Mapped congis | 7 | 4 | 4 |
| Longest contig (AA) | 65 (30.09%) | 110 (50.23%) | 105 (49.07%) |
| Sequence coverage (%) | 211 (97.69%) | 217 (99.09%) | 213 (99.53%) |
| Sequence accuracy (%) | 201 (95.26%) | 205 (94.47%) | 202 (94.84%) |