Table 1.
Specifications of the most commonly used sequencing platforms in microbial community characterization studies.
| Platform | Raw ERa (%) | ER after denoisea (%) | Read length (bp) | Throughput (Gb/run) | Cost/Gb (USD) | Known problems | Reference |
|---|---|---|---|---|---|---|---|
| 454 FLX Titanium | 1.0–2.0 | <0.02 | 450 | 0.4 | 15,500 | High error rate in homopolymer regions. Sequence quality decreases in a lengthwise fashion. Soon to be phased out | (16, 25–28) |
| Illumina MiSeq v2 | 0.8–1.0 | <0.02 | 2 × 250 | 7.5 | 142 | Sequence quality decreases in a lengthwise fashion. The second read has a higher error rate than the first read. Increased single-base errors in association with GGC motifs | (16, 26, 29–31) |
| Ion Torrent PGM 316 chip | 1.5 | NAb | 400 | 1 | 674 | Premature sequence truncation caused by organism- and orientation-dependent biases. Low accuracy in homopolymer regions | (16, 31–33) |
| PacBio RS II | 1.8 | 0.3 | 10,000 | 0.1 | 1,100 | Systematic and non-random errors; G and C are more likely to be deleted than A and C. Preferential loading of shorter sequences into zero-mode waveguides | (16, 27, 31, 34, 35) |
aError rate calculated by sequencing of 16S amplicons from mock bacterial communities.
bTo the best of our knowledge, there are no available studies assessing the error rate of Ion Torrent sequences after bioinformatic curation.