Deep sequencing of evolving pathogen populations: applications, errors, and bioinformatic solutions

Table 2 Sequencing technologies, features and errors

Platform	Manufacturer	Throughput (per machine run)	Reported errors	Depth (virus)	Depth (bacteria)	Reference
454 GS Junior	Roche	~135 K reads @ ~520 nt	~0.38% indels	7 K	14	[26]
GS-FLX Titanium	Roche	~1 M reads @ ~500 nt	~0.28% indels; ~0.12% substitution (max 1.07%)	50 K	100	[27]
MiSeq	Illumina	~ 11 M reads @ ~ 150 nt	< 0.001% indels, ~0.1% substitutions	165 K	330	[26]
GA IIx	Illumina	~ 640 M reads @ 100 nt	~0.001% indels; ~0.31% substitutions (max ~5.85%)	6 M	13 K	[27]
HiSeq 2000	Illumina	~ 6G reads @ 100 nt	~0.002% indels; ~0.32% substitutions (max ~8.2%)	60 M	120 K	*
Ion Torrent PGM	Life technologies	~2 M reads @ ~121 nt	~1.5% indels	24 K	48	[26]
SOLiD	Life technologies	~120 M reads @ ~50 nt	~0.09% substitutions (max > 5%)	600 K	1 K	[28, 29]
RS	Pacific biosystems	~200 K reads @ ~2000 nt (max > 15000 nt)	~14% indels, ~1% substitutions	40 K	80	[30, 31]
tSMS	Helicos	~1G reads @ 35 nt	~3% indels, ~0.2% substitutions	3 M	7 K	[32]

Indels errors are largely associated with homopolymers for Roche and Ion Torrent. This fact can have a significant impact on the detection of variants associated with homopolymers, as was recently shown for the 2184delA mutation of the cystic fibrosis transmembrane conductance regulator (CFTR) using Ion Torrent PGM [33]. Sequencing errors are also highly dependent on the sequencing context and thus can influence variant calling in a biased, but potentially predictable way. For example, certain GC-rich motifs have been reported to have substitution errors of close to 6% [27] for the Illumina sequencing technology. Depth columns give anticipated read depth for a typical viral (~10 K) or bacterial (~5 M) genome.
*Calculated for this review from control PhiX data using GemSIM v1.6 [27].