NGS Data Analysis. Roberto Preste

Transcription

1 NGS Data Analysis Roberto Preste 1

2 Useful info Contacts: Slides: 2

3 NGS data analysis Overview 3

4 NGS Data Analysis: the basic idea 4

5 NGS Data Analysis: the actual workflow Preprocessing & Mapping Variant discovery Functional annotation 5

6 NGS data analysis Quality check & preprocessing 6

7 Fasta files >J Homo sapiens mitochondrion, complete genome GATCACAGGTCTATCACCCTATTAACCACTCACGGGAGCTCTCCATGCATTTGGTATTTTCGTCTGGGGG GTATGCACGCGATAGCATTGCGAGACGCTGGAGCCGGAGCACCCTATGTCGCAGTATCTGTCTTTGATTC CTGCCTCATCCTATTATTTATCGCACCTACGTTCAATATTACAGGCGAACATACTTACTAAAGTGTGTTA ATTAATTAATGCTTGTAGGACATAATAATAACAATTGAATGTCTGCACAGCCACTTTCCACACAGACATC ATAACAAAAAATTTCCACCAAACCCCCCCTCCCCCGCTTCTGGCCACAGCACTTAAACACATCTCTGCCA Both human- and machine-readable Can store multiple sequences ID can contain details or comments Usually contains full genomes or long sequence chunks Sequence ID Sequence 7

8 Fastq Sequence ID GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT + Sequence Spacer!''*((((***+))%%%++)(%%%%).1***-+*''))**55CCF>>>>>>CCCCCC420 Both human- and machine-readable Can store multiple sequences ID can contain sequencing details and technical info Usually contains short sequence chunks (sequencing reads) Quality Score 8

9 Quality score Phred quality score: estimated probability of an error in base calling Probability that the base call is incorrect usually [0-40] Encoded using ASCII characters in fastq files: Quality score Probability of errors ASCII encoding 0-9 1! #$%& ()* /10 +,-./ / :;<=> /10000 I 9

10 Quality GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT +!''*((((***+))%%%++)(%%%%).1***-+*''))**55CCF>>>>>>CCCCCC Quality score Probability of errors ASCII encoding 0-9 1! #$%& ()* /10 +,-./ / :;<=> /1000?@ABCDEFGH 40 1/10000 I

11 Quality check FastQC: visual report of several quality checks for NGS data, useful for further processing Different modules = different checks Pass Warning Fail 11

12 FastQC modules 12

13 Preprocessing Cleansing of reads to solve several issues: Good quality Bad quality Adapter remove adapters 13

14 Preprocessing Cleansing of reads to solve several issues: Good quality Bad quality Adapter remove adapters cut low quality bases from both ends 14

15 Preprocessing Cleansing of reads to solve several issues: Good quality Bad quality Adapter remove adapters cut low quality bases from both ends drop short reads 15

16 Preprocessing Cleansing of reads to solve several issues: remove adapters Common tools: cut low quality bases from both ends drop short reads Trimmomatic Trim Galore! FASTX 16

17 Post-processing quality check Was the processing effective? Are these data ready to be aligned? 17

18 NGS data analysis Alignment 18

19 Alignment vs Assembly Alignment (reference-based) reference genome available reads aligned on it Assembly (de-novo) reference genome not available reads aligned with each other..gtgacttagtcgtagctagctagtagctcgatctaga.. GTGACTTAGT GCTAGCTAGT AGTTAGTCGT GTGACTTAGT GAGTTAGTCG CTTAGTCGTA TAGTCGTAGC TAGTAGCTCG GTAGCTCGAT GTAGCTCGAT TCGTAGCTAG TGAGTTAGCC CGATCTAGA AGCTCGACCT AGCTAGTAGC AGCTCGACCT CTCGACCTAG..GTGACTTAGTCGTAGCTAGCTAGTAGCTCGATCTAGA.. 19

20 Alignment..GTGACTTAGTCGTAGCTAGCTAGTAGCTCGATCTAGA.. GTGACTTAGT GAGTTAGTCG Reference genome TAGTAGCTCG GTAGCTCGAT CTTAGTCGTA TAGTCGTAGC Reads AGCTCGACCT CTCGACCTAG Common aligners: BWA Bowtie GMAP/GSNAP 20

21 Assembly GTGACTTAGT GCTAGCTAGT AGTTAGTCGT CGATCTAGA GTAGCTCGAT Reads TCGTAGCTAG TGAGTTAGCC AGCTCGACCT AGCTAGTAGC..GTGACTTAGTCGTAGCTAGCTAGTAGCTCGATCTAGA.. Common approaches: greedy algorithm graph method Consensus sequence Common assemblers: Newbler SPAdes MaSuRCA 21

22 Paired-end vs single-end reads Suitable for most applications Cheaper Faster Easy to identify read position in genome High accuracy for structural rearrangements and assembly of repetitive regions More expensive 22

23 SAM/BAM files Text-based format used to store aligned reads (to a reference genome) Header Alignments SAM (Sequence Alignment Map) Both human- and machine-readable Header section contains TAG:VALUE pairs Alignment section contains 11 mandatory fields BAM (Binary Alignment Map) Compressed version of SAM Binary format Only machine-readable 23

24 SAM files TAG:VALUE pairs Record type header line reference sequence VN: version number SN: reference sequence name SO: alignments sorting order LN: reference sequence length 24

25 SAM files QNAME MAPQ POS SEQ CIGAR 25

26 Alignment quality check Integrative Genomics Viewer (IGV) Interactive visualization of NGS data from Fasta, SAM, BAM, VCF files Additional features are organized in tracks (gene expression, methylation, copy number variations...) 26

27 NGS data analysis Variant calling 27

28 Variations Aligned data can be used to assess the presence of mutations: somatic germline 28

29 Variations Most common variations searched for: Variation type Reference ACTGACGCATGCATCATGCATGC SNP ACTGACGCATGCATCATTCATGC Insertion ACTGACGCATGGTACATCATGCATGC Deletion ACTGACGC--GCATCATGCATGC Variation effect 29

30 Variant callers A plethora of different tools, each with its own peculiarities: variation type (SNV vs structural variation) source (somatic vs cancer mutations) only detect variants vs also predict their effect Variant Call Format (VCF) 30

31 VCF files Header (information and metadata) Variants Variant annotations Genotype annotations 31

32 VCF files Mandatory fields: ALT: alternative base(s) #CHROM: chromosome or contig QUAL: Phred quality score for each ALT POS: variant position (1-based) FILTER: variant call filter status ID: variant identifier (usually dbsnp ID) REF: reference base(s) INFO: key-value pairs with additional information (described in header) Optional fields: FORMAT: genotype information fields SAMPLE1 SAMPLEn: values for fields listed in FORMAT 32

33 NGS data analysis Functional annotation 33

34 Functional annotation small variants (SNP/indels) per genome ATCATGCATGC ATCATTCATGC Which ones are most interesting for our purpose? 34

35 Annovar Identify variants and flag those with detrimental effects Gene-based annotations identify variants that can cause protein coding changes detect the amino acids that are affected 35

36 Annovar Identify variants and flag those with detrimental effects Region-based annotations identify variants in specific genomic regions: conserved regions predicted transcription factor binding sites segmental duplication regions etc 36

37 Annovar Identify variants and flag those with detrimental effects Filter-based annotations identify variants that are documented in specific databases: dbsnp 1000 Genome Project etc 37

38 Annovar VCF files Extensive number of new annotations added to the initial VCF file 38

39 MToolBox Human mtdna reconstruction, analysis and annotation from NGS data Haplogroup predictions Both command-line and web-based versions available 39

40 HmtDB Over human mitochondrial sequences Healthy/patient and continent-specific subsets Genome-centric 40

41 HmtVar Over human mitochondrial variants Pathogenicity predictions available for most variants Variant-centric 41

42 NGS Data Analysis: pipelines General-purpose programming language Easy to learn Very powerful (libraries available for anything you can think of) Biopython: specific module for bioinformatics 42

43 NGS Data Analysis: pipelines Originally suited for statistics Particularly used for data analysis and visualization Gained a lot of traction for many different disciplines Bioconductor: specific package for bioinformatics 43

44 Useful info Contacts: Slides: 44