Table 1.
Performance comparison of assembly software packages
| Metagenome dataset | Separate assembly | MetaGenomic assembly |
|||
|---|---|---|---|---|---|
| Velvet | MetaVelvet | Velvet | SOAPdenovo | Meta-IDBA | |
| Order level (total genome size = 71 929 175 bp; 93 423 332 reads) | |||||
| Num. scaffolds | 685 | 813 | 924 | 5 998 | 2 678 |
| Total scaffold length | 71 009 045 | 71 053 228 | 48 450 203 | 70 296 665 | 70 312 381 |
| N50 size (bp) | 288 838 | 268 350 | 142 471 | 43 796 | 55 575 |
| Chimeric scaffold length (%) | 0.00 | 0.00 | 0.46 | 0.00 | 0.00 |
| Cover rate (%) | 98.38 | 98.25 | 67.48 | 95.67 | 96.98 |
| Number of predicted genes | 66 268 | 66 241 | 43 729 | 60 319 | 62 833 |
| Required CPU time (s) | 4 994 | 8 685 | 7 076 | 11 564 | 7 375 |
| Required memory (GB) | 7.04 | 56.61 | 54.07 | 62.42 | 15.15 |
| Family level (total genome size = 84 552 832 bp; 113 680 114 reads) | |||||
| Num. scaffolds | 784 | 1 019 | 2 889 | 9 039 | 4 421 |
| Total scaffold length | 83 275 357 | 83 322 440 | 65 789 192 | 81 739 588 | 81 990 799 |
| N50 size (bp) | 313 454 | 257 853 | 76 239 | 27 510 | 39 961 |
| Chimeric scaffold length (%) | 0.00 | 0.45 | 0.02 | 0.03 | 0.00 |
| Cover rate (%) | 98.12 | 97.81 | 77.60 | 94.09 | 96.03 |
| Number of predicted genes | 77 634 | 77 655 | 58 744 | 68 832 | 72 746 |
| Required CPU time (s) | 9585 | 11 409 | 9 813 | 14 803 | 12 664 |
| Required memory (GB) | 13.15 | 72.06 | 68.98 | 62.48 | 23.11 |
| Genus level (total genome size = 88 595 850 bp; 103 990 387 reads) | |||||
| Num. scaffolds | 1288 | 2 325 | 3 633 | 10 282 | 10 643 |
| Total scaffold length | 86 489 808 | 84 342 495 | 53 450 902 | 79 334 848 | 74 808 521 |
| N50 size (bp) | 279 359 | 239 061 | 74 182 | 16 194 | 12 773 |
| Chimeric scaffold length (%) | 0.00 | 1.56 | 0.00 | 0.08 | 0.00 |
| Cover rate (%) | 98.17 | 97.13 | 73.31 | 91.73 | 90.93 |
| Number of predicted genes | 80 812 | 79 301 | 46 688 | 67 267 | 61 135 |
| Required CPU time (s) | 7275 | 10 395 | 8 712 | 12 889 | 15 071 |
| Required memory (GB) | 11.12 | 63.22 | 60.43 | 62.45 | 16.75 |
| Species level (total genome size = 85 450 435 bp; 98 817 303 reads) | |||||
| Num. scaffolds | 818 | 3 447 | 2 403 | 9 317 | 6 657 |
| Total scaffold length | 83 865 679 | 80 628 784 | 40 619 181 | 70 762 160 | 64 880 992 |
| N50 size (bp) | 339 109 | 152 531 | 100 819 | 14 471 | 26 571 |
| Chimeric scaffold length (%) | 0.00 | 0.93 | 0.00 | 0.01 | 0.00 |
| Cover rate (%) | 97.79 | 94.56 | 60.29 | 84.62 | 82.50 |
| Number of predicted genes | 83 952 | 81 842 | 38 445 | 65 176 | 58 367 |
| Required CPU time (s) | 7618 | 12 001 | 8 775 | 12 858 | 20 755 |
| Required memory (GB) | 7.68 | 64.06 | 61.23 | 62.46 | 17.32 |
All computations were executed with Intel(R) Xeon(R) E5540 processors (2.53 GHz), with 48 GB physical memory, except for a few cases. The figures in ‘separate assembly’ show the results of single-genome assembly from pure sequence reads of each single-isolate genome, which were not available in real-data analysis. MetaVelvet, Velvet and SOAPdenovo were run with default parameters, except for setting k-mer size at 51. Meta-IDBA was run with default parameters, except for setting the maximum k-mer size at 50.