Welcome to the blog

Posts

My thoughts and ideas

Alignment Visualization | Griffith Lab

RNA-seq Bioinformatics

Introduction to bioinformatics for RNA sequence analysis

Alignment Visualization


RNA-seq_Flowchart3


Before we can view our alignments in the IGV browser we need to index our BAM files. We will use samtools index for this purpose. For convenience later, index all bam files.

    echo $RNA_ALIGN_DIR
    cd $RNA_ALIGN_DIR
    find *.bam -exec echo samtools index {} \; | sh

Visualize alignments

Start IGV on your laptop. Load the UHR.bam & HBR.bam files in IGV. You can load the necessary files in IGV directly from your web accessible amazon workspace (see below) using ‘File’ -> ‘Load from URL’. You may wish to customize the track names as you load them in to keep them straight. Do this by right-clicking on the alignment track and choosing ‘Rename Track’.

UHR hisat2 alignment:

http://YOUR_DNS_NAME/workspace/rnaseq/alignments/hisat2/UHR.bam

HBR hisat2 alignment:

http://YOUR_DNS_NAME/workspace/rnaseq/alignments/hisat2/HBR.bam

Go to an example gene locus on chr22:

  • e.g. EIF3L, NDUFA6, and RBX1 have nice coverage
  • e.g. SULT4A1 and GTSE1 are differentially expressed. Are they up-regulated or down-regulated in the brain (HBR) compared to cancer cell lines (UHR)?
  • Mouse over some reads and use the read group (RG) flag to determine which replicate the reads come from. What other details can you learn about each read and its alignment to the reference genome.

Exercise

Try to find a variant position in the RNAseq data:

  • HINT: DDX17 is a highly expressed gene with several variants in its 3 prime UTR.
  • Other highly expressed genes you might explore are: NUP50, CYB5R3, and EIF3L (all have at least one transcribed variant).
  • Are these variants previously known (e.g., present in dbSNP)?
  • How should we interpret the allele frequency of each variant? Remember that we have rather unusual samples here in that they are actually pooled RNAs corresponding to multiple individuals (genotypes).
  • Take note of the genomic position of your variant. We will need this later.

BAM Read Counting

Using one of the variant positions identified above, count the number of supporting reference and variant reads. First, use samtools mpileup to visualize a region of alignment with a variant.

    cd $RNA_HOME
    mkdir bam_readcount
    cd bam_readcount

Create faidx indexed reference sequence file for use with mpileup

    echo $RNA_REF_FASTA
    samtools faidx $RNA_REF_FASTA

Run samtools mpileup on a region of interest

    samtools mpileup -f $RNA_REF_FASTA -r 22:18918457-18918467 $RNA_ALIGN_DIR/UHR.bam $RNA_ALIGN_DIR/HBR.bam

Each line consists of chromosome, 1-based coordinate, reference base, the number of reads covering the site, read bases and base qualities. At the read base column, a dot stands for a match to the reference base on the forward strand, a comma for a match on the reverse strand, ACGTN for a mismatch on the forward strand and acgtn for a mismatch on the reverse strand. A pattern \+[0-9]+[ACGTNacgtn]+ indicates there is an insertion between this reference position and the next reference position. The length of the insertion is given by the integer in the pattern, followed by the inserted sequence. See samtools pileup/mpileup documentation for more explanation of the output:

Now, use bam-readcount to count reference and variant bases at a specific position. First, create a bed file with some positions of interest (we will create a file called snvs.bed using the echo command).

It will contain a single line specifying a variant position on chr22 e.g.:

22 38483683 38483683

Create the bed file

    echo "22 38483683 38483683"
    echo "22 38483683 38483683" > snvs.bed

Run bam-readcount on this list for the tumor and normal merged bam files

    bam-readcount -l snvs.bed -f $RNA_REF_FASTA $RNA_ALIGN_DIR/UHR.bam 2>/dev/null
    bam-readcount -l snvs.bed -f $RNA_REF_FASTA $RNA_ALIGN_DIR/HBR.bam 2>/dev/null

Now, run it again, but ignore stderr and redirect stdout to a file:

    bam-readcount -l snvs.bed -f $RNA_REF_FASTA $RNA_ALIGN_DIR/UHR.bam 2>/dev/null 1>UHR_bam-readcounts.txt
    bam-readcount -l snvs.bed -f $RNA_REF_FASTA $RNA_ALIGN_DIR/HBR.bam 2>/dev/null 1>HBR_bam-readcounts.txt

From this output you could parse the read counts for each base

    cat UHR_bam-readcounts.txt | perl -ne '@data=split("\t", $_); @Adata=split(":", $data[5]); @Cdata=split(":", $data[6]); @Gdata=split(":", $data[7]); @Tdata=split(":", $data[8]); print "UHR Counts\t$data[0]\t$data[1]\tA: $Adata[1]\tC: $Cdata[1]\tT: $Tdata[1]\tG: $Gdata[1]\n";'
    cat HBR_bam-readcounts.txt | perl -ne '@data=split("\t", $_); @Adata=split(":", $data[5]); @Cdata=split(":", $data[6]); @Gdata=split(":", $data[7]); @Tdata=split(":", $data[8]); print "HBR Counts\t$data[0]\t$data[1]\tA: $Adata[1]\tC: $Cdata[1]\tT: $Tdata[1]\tG: $Gdata[1]\n";'

PRACTICAL EXERCISE 7

Assignment: Index your bam files from Practical Exercise 6 and visualize in IGV.

  • Hint: As before, it may be simplest to just index and visualize the combined/merged bam files HCC1395_normal.bam and HCC1395_tumor.bam.
  • If this works, you should have two BAM files that can be loaded into IGV from the following location on your cloud instance:
    • http://YOUR_DNS_NAME/workspace/rnaseq/practice/alignments/hisat2/

Questions

  • Load your merged normal and tumor BAM files into IGV. Navigate to this location on chromosome 22: ‘chr22:38,466,394-38,508,115’. What do you see here? How would you describe the direction of transcription for the two genes? Does the reported strand for the reads aligned to each of these genes appear to make sense? How do you modify IGV settings to see the strand clearly?
  • How can we modify IGV to color reads by Read Group? How many read groups are there for each sample (tumor & normal)? What are your read group names for the tumor sample?
  • What are the options for visualizing splicing or alternative splicing patterns in IGV? Navigate to this location on chromosome 22: ‘chr22:40,363,200-40,367,500’. What splicing event do you see?

Solution: When you are ready you can check your approach against the Solutions.

Alignment | Griffith Lab

RNA-seq Bioinformatics

Introduction to bioinformatics for RNA sequence analysis

Alignment

HISAT2 alignment

Perform alignments with HISAT2 to the genome and transcriptome.

First, begin by making the appropriate output directory for our alignment results.

    echo $RNA_ALIGN_DIR
    mkdir -p $RNA_ALIGN_DIR
    cd $RNA_ALIGN_DIR

HISAT2 uses a graph-based alignment and has succeeded HISAT and TOPHAT2. The output of this step will be a SAM/BAM file for each data set.

Refer to HISAT2 manual for a more detailed explanation:

HISAT2 basic usage:

    #hisat2 [options]* -x <ht2-idx> {-1 <m1> -2 <m2> | -U <r> | --sra-acc <SRA accession number>} [-S <sam>]

Extra options specified below:

hisat2 -p 8 --rg-id=UHR_Rep1 --rg SM:UHR --rg LB:UHR_Rep1_ERCC-Mix1 --rg PL:ILLUMINA --rg PU:CXX1234-ACTGAC.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/UHR_Rep1_ERCC-Mix1_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/UHR_Rep1_ERCC-Mix1_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./UHR_Rep1.sam
hisat2 -p 8 --rg-id=UHR_Rep2 --rg SM:UHR --rg LB:UHR_Rep2_ERCC-Mix1 --rg PL:ILLUMINA --rg PU:CXX1234-TGACAC.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/UHR_Rep2_ERCC-Mix1_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/UHR_Rep2_ERCC-Mix1_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./UHR_Rep2.sam
hisat2 -p 8 --rg-id=UHR_Rep3 --rg SM:UHR --rg LB:UHR_Rep3_ERCC-Mix1 --rg PL:ILLUMINA --rg PU:CXX1234-CTGACA.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/UHR_Rep3_ERCC-Mix1_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/UHR_Rep3_ERCC-Mix1_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./UHR_Rep3.sam

hisat2 -p 8 --rg-id=HBR_Rep1 --rg SM:HBR --rg LB:HBR_Rep1_ERCC-Mix2 --rg PL:ILLUMINA --rg PU:CXX1234-TGACAC.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/HBR_Rep1_ERCC-Mix2_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/HBR_Rep1_ERCC-Mix2_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./HBR_Rep1.sam
hisat2 -p 8 --rg-id=HBR_Rep2 --rg SM:HBR --rg LB:HBR_Rep2_ERCC-Mix2 --rg PL:ILLUMINA --rg PU:CXX1234-GACACT.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/HBR_Rep2_ERCC-Mix2_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/HBR_Rep2_ERCC-Mix2_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./HBR_Rep2.sam
hisat2 -p 8 --rg-id=HBR_Rep3 --rg SM:HBR --rg LB:HBR_Rep3_ERCC-Mix2 --rg PL:ILLUMINA --rg PU:CXX1234-ACACTG.1 -x $RNA_REF_INDEX --dta --rna-strandness RF -1 $RNA_DATA_DIR/HBR_Rep3_ERCC-Mix2_Build37-ErccTranscripts-chr22.read1.fastq.gz -2 $RNA_DATA_DIR/HBR_Rep3_ERCC-Mix2_Build37-ErccTranscripts-chr22.read2.fastq.gz -S ./HBR_Rep3.sam

Note: in the above alignments, we are treating each library as an independent data set. If you had multiple lanes of data for a single library, you could align them all together in one HISAT2 command. Similarly you might combine technical replicates into a single alignment run (perhaps after examining them and removing outliers…). To combine multiple lanes, you would provide all the read1 files as a comma separated list for the ‘-1’ input argument, and then all read2 files as a comma separated list for the ‘-2’ input argument, (where both lists have the same order) : You can also use samtools merge to combine bam files after alignment. This is the approach we will take.

HISAT2 Alignment Summary HISAT2 generates a summary of the alignments printed to the terminal. Notice the number of total reads, reads aligned and various metrics regarding how the reads aligned to the reference.

SAM to BAM Conversion Convert HISAT2 sam files to bam files and sort by aligned position

    samtools sort -@ 8 -o UHR_Rep1.bam UHR_Rep1.sam
    samtools sort -@ 8 -o UHR_Rep2.bam UHR_Rep2.sam
    samtools sort -@ 8 -o UHR_Rep3.bam UHR_Rep3.sam
    samtools sort -@ 8 -o HBR_Rep1.bam HBR_Rep1.sam
    samtools sort -@ 8 -o HBR_Rep2.bam HBR_Rep2.sam
    samtools sort -@ 8 -o HBR_Rep3.bam HBR_Rep3.sam

Merge HISAT2 BAM files

Make a single BAM file combining all UHR data and another for all HBR data. Note: This could be done in several ways such as ‘samtools merge’, ‘bamtools merge’, or using picard-tools (see below). We chose the third method because it did the best job at merging the bam header information. NOTE: sambamba also retains header info.

    cd $RNA_HOME/alignments/hisat2
    java -Xmx2g -jar $RNA_HOME/tools/picard.jar MergeSamFiles OUTPUT=UHR.bam INPUT=UHR_Rep1.bam INPUT=UHR_Rep2.bam INPUT=UHR_Rep3.bam
    java -Xmx2g -jar $RNA_HOME/tools/picard.jar MergeSamFiles OUTPUT=HBR.bam INPUT=HBR_Rep1.bam INPUT=HBR_Rep2.bam INPUT=HBR_Rep3.bam

Count the alignment (BAM) files to make sure all were created successfully (you should have 8 total)

    ls -l *.bam | wc -l
    ls -l *.bam

PRACTICAL EXERCISE 6

Assignment: Perform some alignments on additional read data sets. Align the reads using the skills you learned above. Try using the HISAT2 aligner. Also practice converting SAM to BAM files, and merging BAM files.

Hint: Do this analysis on the additional data and in the separate working directory called ‘practice’ that you created in Practical Exercise 3. Questions

What is the difference between a .sam and .bam file? If you sorted the resulting BAM file as we did above, is the result sorted by read name? Or position? Which columns of the BAM file can be viewed to determine the style of sorting? What command can you use to view only the BAM header?

Solution: When you are ready you can check your approach against the Solutions.