2013 NGS Field Guide – Table 3c – Error Rates

Table 3c. Error rates for announced & commercially available DNA sequencing platforms. All Error rates are in percentage. Percentage of errors per base within single reads of the maximum length. As explained in Glenn (2011), error rates among platforms are not exactly comparable. The reported Ion Torrent rates range from 0.46% to 2.4%. Final Error rates SOLiD rates are from reads with bases consistent on double or triple sequencing only. Final Error rate for PacBio applies only to the consensus sequencing for three independent reads of the same template. For Illumina, errors at ≤0.1% is achieved for ≥ 85% of reads (not all reads). See Glenn (2011) for additional details.


Primary Errors Single-pass Error Rate (%) Final Error Rate (%)
3730xl (capillary) Substitution 0.1-1 0.1-1
454, all models Indel 1 1
Illumina, all models Substitution ~0.1 ~0.1
Ion Torrent – all chips Indel ~1 ~1
SOLiD – 5500xl A-T bias ~5 ≤ 0.1
Oxford Nanopore deletions ≥ 4* 4*
PacBio RS indel ~13 ≤ 1

* Information based on company sources alone (independent data not yet available); it is not clear if the 4% error rate reported by Oxford Nanopore refers to a single-pass rate or is what is achieved after reading both strands and producing a consensus sequence.

  • David

    Ya, hard to compare. The final error rate for PacBio runs using their Quiver assembler is better than Q50 (99.999%) and you’ll find publications recently at Q60 (99.9999%). That’s without circular consensus reads, just consensus from raw reads. Related to the lack of bias perhaps: http://genomebiology.com/2013/14/5/R51