High Throughput Sequencing
Xiaole Shirley Liu
STAT115, STAT215, BIO298, BIST520
First Generation
• Sanger Sequencing: sequencing and detection 2 different steps: 384 * 1kb / 3 hours
2
Second Generation
• Massively parallel sequencing by synthesis
• Many different technologies: Illumina, 454, SOLiD, Helicos, etc
• Illumina: HiSeq, MiSeq, NextSeq• 1-16 samples • 25M-4B reads• 30-300bp • 1-8 days• 15GB-1TB output• Moving targets
3
Illumina Cluster Generation
• Amplify sequenced fragments in place on the flow cell
• Can sequence from both the pink and purple adapters (Paired-end seq)
• Can multiplex many samples / lane
4
Illumina Sequencing
5
Third Generation• Single molecule sequencing: no amp• Fewer but much longer reads• Good for sequencing long reads, but not for read count
applications, technology still in developmenthttp://www.youtube.com/watch?v=v8p4ph2MAvI
• https://www.nanoporetech.com/news/movies#movie-28-minion
6
High Throughput Sequencing
• Big (data), fast (speed), cheap (cost), flexible (applications)
• Bioinformatic analyses become bottleneck
7
High Throughput Sequencing Data Analysis
8
FASTQ File
• Format– Sequence ID, sequence
– Quality ID, quality score
• Quality score using ASCII (higher -> better)
9
@HWI-EAS305:1:1:1:991#0/1
GCTGGAGGTTCAGGCTGGCCGGATTTAAACGTAT
+HWI-EAS305:1:1:1:991#0/1
MVXUWVRKTWWULRQQMMWWBBBBBBBBBBBBBB
@HWI-EAS305:1:1:1:201#0/1
AAGACAAAGATGTGCTTTCTAAATCTGCACTAAT
+HWI-EAS305:1:1:1:201#0/1
PXX[[[[XTXYXTTWYYY[XXWWW[TMTVXWBBB
FASTQC: Sequencing Quality
10
Read Mapping
• Mapping hundreds of millions of reads back to the reference genome is CPU and RAM intensive and slow
• Read quality decreases with length (small single nucleotide mismatches or indels)
• Most mappers allow ~2 mismatches within first 30bp (4 ^ 28 could still uniquely identify most 30bp sequences in a 3GB genome), slower when allowing indels
• Mapping output: SAM (BAM) or BED11
Spaced seed alignment
• Tags and tag-sized pieces of reference are cut into small “seeds.”
• Pairs of spaced seeds are stored in an index.
• Look up spaced seeds for each tag.
• For each “hit,” confirm the remaining positions.
• Report results to the user.
Burrows-Wheeler
• Store entire reference genome.
• Align tag base by base from the end.
• When tag is traversed, all active locations are reported.
• If no match is found, then back up and try a substitution.
Trapnell & Salzberg, Nat Biotech 2009
Burrows-Wheeler Transform
• Reversible permutation used originally in compression
• Once BWT(T) is built, all else shown here is discarded– Matrix will be shown for illustration only
BurrowsWheelerMatrix
Last column
BWT(T)T
Burrows M, Wheeler DJ: A block sorting lossless data compression algorithm. Digital Equipment Corporation, Palo Alto, CA 1994, Technical Report 124; 1994Slides from Ben Langmead
Encoding for compressiongc$ac1111001
Burrows-Wheeler Transform
• Property that makes BWT(T) reversible is “LF Mapping”– ith occurrence of a character in Last column is
same text occurrence as the ith occurrence in First column
T
BWT(T)
Burrows WheelerMatrix
Rank: 2
Rank: 2
Slides from Ben Langmead
Burrows-Wheeler Transform
• To recreate T from BWT(T), repeatedly apply rule:T = BWT[ LF(i) ] + T; i = LF(i)– Where LF(i) maps row i to row whose first
character corresponds to i’s last per LF Mapping Final T
Slides from Ben Langmead
Exact Matching with FM Index
• To match Q in T using BWT(T), repeatedly apply rule:top = LF(top, qc); bot = LF(bot, qc)– Where qc is the next character in Q (right-to-
left) and LF(i, qc) maps row i to the row whose first character corresponds to i’s last character as if it were qc
Slides from Ben Langmead
Exact Matching with FM Index
• In progressive rounds, top & bot delimit the range of rows beginning with progressively longer suffixes of Q (from right to left)
• If range becomes empty the query suffix (and therefore the query) does not occur in the text
• If no match, instead of giving up, try to “backtrack” to a previous position and try a different base (mismatch, much slower)
Slides from Ben Langmead
STAR Alignment
• Suffix Tree
• Very fast and accuracy for mapping PE-seq and high read counts
• O(n) time to build
• O(mlogn) time to
search
19
Suffix tree (Example)
Let s=abab, a suffix tree of s is a compressed trie of all suffixes of s=abab$
{ $ b$ ab$ bab$ abab$ }
ab
ab
$
ab$
b
$
$
$
Mapped Seq Files
• Mapped SAM
– Map: 0 OK, 4 unmapped, 16 mapped reverse strand
– Sequence, quality score
– XA (mapper-specific)
– MD: mismatch info: 3 match, then C ref, 30 match, then T ref, 3 match
– NM: number of mismatch
• BAM: binary SAM format
• Mapped BED
– Chr, start, end, strand
21
HWUSI-EAS366_0112:6:1:1298:18828#0/1 16 chr9 98116600 255 38M * 0 0 TACAATATGTCTTTATTTGAGATATGGATTTTAGGCCG Y\]bc^dab\[_UU`^`LbTUT\ccLbbYaY`cWLYW^ XA:i:1 MD:Z:3C30T3 NM:i:2
HWUSI-EAS366_0112:6:1:1257:18819#0/1 4 * 0 0 * * 0 0 AGACCACATGAAGCTCAAGAAGAAGGAAGACAAAAGTG ece^dddT\cT^c`a`ccdK\c^^__]Yb\_cKS^_W\ XM:i:1
HWUSI-EAS366_0112:6:1:1315:19529#0/1 16 chr9 102610263 255 38M * 0 0 GCACTCAAGGGTACAGGAAAAGGGTCAGAAGTGTGGCC ^c_Yc\Lcb`bbYdTa\dd\`dda`cdd\Y\ddd^cT` XA:i:0 MD:Z:38 NM:i:0
chr1 123450 123500 +chr5 2837461528374615-http://samtools.github.io/hts-specs/SAMv1.pdf
Mapping Statistics Terms
• Mappable locations: reads that can find match to A location in the genome
• Uniquely mapped reads: reads that can find match to A SINGLE location in the genome– Repeat sequences in the genome, length-
dependent
• Uniquely mapped locations: number of unique locations hit by uniquely mapped reads– Redundancy: potential PCR amplification bias
22
Summary
• Sequencing technologies– 1st, 2nd, 3rd generation
• Sequence quality assessment– FASTQC
• Read mapping– Spaced seed
– BWA: Borrows Wheeler transformation, LF mapping
– STAR: Suffix Tree, fast
• SAM / BAM format
23
