Scribd
Upload
12 views

Lecture 03

Uploaded by

Sourav Kumar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
12 views

Lecture 03

Uploaded by

Sourav Kumar
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 36

Next Generation Sequencing (NGS)

Lecture 03
Amit Ghosh
IIT Kharagpur

http://www.energy.iitkgp.ac.in/~amitghosh/

School of Energy Science & Engineering


Technical Terms
ATGCAGAGAGTCGA…….
Read

ATGCAGAGAGTCGA…….
Library CAGAGGCTACGGATGC…….

AGTGATAGCTATGACA…….

Single-end sequencing

Paired-end sequencing

School of Energy Science & Engineering


NGS Data Processing

Sequencing depth or coverage

Reads

Quality score or PHRED Score (Q)

P(Error)=10(-Q/10)
Q = 40, P(Error) = 10-4
Q = 10, P(Error) = 10-1

Percentage Accuracy = [1 - P(Error)] x 100

School of Energy Science & Engineering


FASTQ Files

• Format
1. Sequence ID
2. Sequence
3. Quality ID
4. Quality Score

School of Energy Science & Engineering


Quality score –ASCII table

School of Energy Science & Engineering


Checking NGS Data Quality

FastQC

https://www.bioinformatics.babraham.ac.uk/projects/fastqc/

School of Energy Science & Engineering


Per base quality score

Good Data

School of Energy Science & Engineering


Per base quality score

Bad Data

School of Energy Science & Engineering


FASTQC: Per sequence quality scores

Good Data

School of Energy Science & Engineering


FASTQC: Per sequence quality scores

Not so good

School of Energy Science & Engineering


FASTQC: Nucleotide Content Per Position

Good Quality Poor Quality

Smooth over length Sequence – position bias

School of Energy Science & Engineering


FASTQC: Per sequence GC content

Good Quality Poor Quality

Fits with expectation Does not fit with expectation

School of Energy Science & Engineering


Data Processing

1. Read mapping against available reference genome sequence

School of Energy Science & Engineering


Data Processing

2. De novo genome assembly

School of Energy Science & Engineering


Mapping read data
Challenges

• Long reference sequences


• Large number of reads
• Reads are small
• Normal algorithms will take too much time (in years!)

Algorithms
if one mapping takes 0.1 sec, mapping 100
Trivial search (slowest) million reads will take –
Blast etc. 0.1x100x106= 106 seconds = 11.5 days
Hash-table based
Suffix-tree based
-large memory requirement
School of Energy Science & Engineering
Burrows-Wheeler Aligners
Most widely used tools:
bwa: http://bio-bwa.sourceforge.net/
Bowtie: http://bowtie-bio.sourceforge.net/index.shtml

BWA

Bowtie

School of Energy Science & Engineering


Burrows-Wheeler transformation

Step 1: Add $at the end and $<a lexicographically

Reference string T: acaacg$

Step 2: Rotate the string counter-clockwise and get all


possible rotation

Step 3: Sort alphabetically and store the last column

School of Energy Science & Engineering


Last-First (LF) mapping
i-th instance of a letter in the last column corresponds to the i-th
instance of the same letter in the first column

School of Energy Science & Engineering


Burrows-Wheeler transformation

Index the reference sequence so that searching is efficient


Reversible permutation used originally in compression

BWT(T) =

Burrows Last column


Wheeler
Matrix

Burrows, M; Wheeler, DJ. A block sorting lossless data compression algorithm, Digital
School of Energy
Equipment Science
Corporation, & Engineering
Palo Alto, CA 1994, Technical Report 124, 1994
Last-First (LF) mapping

i-th instance of a letter in the last column corresponds to the i-th


instance of the same letter in the first column

School of Energy Science & Engineering


BWT(T) to Retrieve Alignments

Query Q = aac

School of Energy Science & Engineering


BWT(T) to Retrieve Alignments

• In progressive rounds, top & bottom delimit the range of rows


beginning

• If range becomes empty 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)

School of Energy Science & Engineering


Mapping a substring in the reference string

School of Energy Science & Engineering


BWT(T) to Retrieve Alignments
• How to recover the query sequence (Q) alignment position in
the reference sequence T: LF mapping

School of Energy Science & Engineering


Alignment output: SAM File - Header

School of Energy Science & Engineering


Mapped Seq Files in SAM format

• Read Name
• Map: 0 OK, 4 unmapped, 16 mapped
• Sequence, quality score
• MD: mismatch info: 3 match, then C ref, 30 match, then T ref, 3 match
• NM: number of mismatch
• BAM: binary compressed SAM format

School of Energy Science & Engineering


Single nucleotide polymorphisms (SNP)

School of Energy Science & Engineering


SNP call with number of reads and quality cut-off

Quality cut-off for SNP base: Q

Cut-off for number of reads showing the reads: C

School of Energy Science & Engineering


SNP call with number of reads and quality cut-off

Cut-off for number of reads showing the reads: C

Quality cut-off for SNP base: Q

How does it help?

Higher confidence in the SNP call and reduced number of false positives

Prob. of error/false call = 10-Q/10x 10-Q/10…. x 10-Q/10(C times)


= 10-cQ/10

School of Energy Science & Engineering


Frequency of SNP

School of Energy Science & Engineering


Copy number variations (CNVs)

Comparison between treatment vs control group

Diseased vs healthy

Cancer vs Normal

Coverage in diseased sample


Coverage ratio (CR) = --------------------------------------------
Coverage in healthy sample

School of Energy Science & Engineering


Coverage ratio (CR)

Coverage in diseased sample


Coverage ratio (CR) = --------------------------------------------
Coverage in healthy sample

Region Coverage in Coverage in


diseased healthy
Region 1 100 50
Region 2 100 100
Region 3 50 150

School of Energy Science & Engineering


Coverage ratio (CR)

Coverage in diseased sample


Coverage ratio (CR) = --------------------------------------------
Coverage in healthy sample

The total coverage might vary across samples

(Coverage in diseased sample)/(Total no of reads in diseased sample)


Coverage ratio (CR) = -------------------------------------------------------------------------
(Coverage in healthy sample)/(Total no of reads in healthy sample)

School of Energy Science & Engineering


Segmentation algorithm

Coverage ratio (CR) values in a genomic region

School of Energy Science & Engineering


Segmentation algorithm

Coverage ratio (CR) values in a genomic region

Multiple windows each with CR ≥ CRthr

School of Energy Science & Engineering


Summary

• Sequencing technologies: 1st, 2nd, 3rd generation


• Illumina (2nd gen) has taken most of the market
• Sequences are sorted in FASTQ file
• After sequencing, perform quality assessment (FASTQC)
• Sequenced “reads” need to be aligned back to reference genome
• BLAST
• Suffix Array
• BWA/Bowtie: Burrows-Wheeler transformation, LF mapping
• Aligned reads are stored in SAM/BAM files

School of Energy Science & Engineering

You might also like