brownbag-science

Types of NGS

	Sanger	MiSeq	HiSeq	NovaSeq
Reads (millions) from a single run	0.0004	30	3,000	13,000
Gigabases/day	0.001	7	500	4000

also see cost of sequencing

		Illumina
	Sanger	MiSeq	NextSeq	HiSeq	NovaSeq	Ion Torrent	PacificBiosciences	OxfordNanopore
Throughput range per run (Gb)	c. 0.0005	10–15	10–120	1000–1800	2000–6000	1–15	0.5–10	0.1–1
Read length	Up to 1 kb	300	150	150	250	200–400	up to 60 kb	up to 100 kb
Read type	SR	SR, PE	SR, PE	SR, PE	SR, PE	SR	SR	SR
Error rate (%)	0.001–1	0.8	0.8	0.2	0.2–0.8	1–2	13	5–40
Error type	Substitutions	Substitutions	Substitutions	Substitutions	Substitutions	Indels, homopolymers	Indels	Indels, deletions
Advantages	Read accuracy and length	Read length	Throughput	Throughput, low error rate	High throughput	Speed, read length	Speed, read length	Read length, portability

also see another comparision

	Advantages	Limitations	Examples
Short-Read	· Higher sequence fidelity · Cheap · Can sequence fragmented DNA	· Not able to resolve structural variants, phasing alleles or distinguish highly homologous genomic regions · Unable to provide coverage of some repetitive regions	* Ion Torrent * 454 * Illumina * SOLiD * cPAS * MPSS
Long-Read	· Able to sequence genetic regions that are difficult to characterize with short-read seq due to repeat sequences · Able to resolve structural rearrangements or homologous regions · Able to read through an entire RNA transcript to determine the specific isoform · Assists de novo genome assembly	· Lower per read accuracy · Bioinformatic challenges, caused by coverage biases, high error rates in base allocation, scalability and limited availability of appropriate pipelines	Pac-Bio (2nd) * Single molecule real time sequencing (3rd Gen) Nanopore (3rd Gen)

(adapted from technologynetworks.com, wikipedia.org)

Also:

• Whole-genome vs Whole-exome sequencing

• Whole-exome (DNA sequencing) vs RNA-Seq (RNA sequencing)

---

Illumina Sequencing

• 90% of sequencing market
• millions to billions of reads per run
• 300 to 600 bases per read
• high fidelity > 99.9% accuracy
• can do $1000 genome in 48 hours
• Issues/limitations
  • length limits - limited to 300 bases (errors from chemistry adds up)
  • some lagging, some jumping ahead
  • sometimes hard to tell clusters apart
  • able to detect differences in clusters when differences occur, look in middle
  • colors can be similar (blue, green, yellow) close to each other in frequency

II. Long-Read Sequencing (Nanopore)

• single strand of dna is threaded through pore, current is measured

• high error rates (10-15%)

• • errors are biased

• can read RNA directly

• really long reads (2 Mb)

• extremely portable (can be done in field), very small device!

• See Nature Review Drug Discovery vol 1, p. 77-84 (2002)

III. PacBio (also synthesis based sequencer)

different chemistry

• no blocker
• fluorescent group is on phosphate group)
• 4 bases, four colors
• camera monitors image in real time
• 1 to 8 million wells
• long reads (100kb)
• High error rates (10-15%)
  • errors are random! (good)
  • if you sample same fragment several times, errors go away with consensus sequence
  • can yield accuracy of 1:1000 to 1:10,000 (which exceeds illumina sequencing accuracy)
• nature review genetics 11, 31-46 (2010)

CONS:

• harder to prepare
• cost more

PROS

• if you want to assemble new genome (no reference sequence), MUCH easier

• • (sometimes you get structural variants in short reads, sequences are inverted)

• much better at structural variation detection (important for certain cancers)

• phase variation: which variant are on which chromosome?

• • can be difficult to determine if mutation is on chromosome A or chromosome B (we get two)