Bottlenose dolphin population genomic analysis

From wiki
Revision as of 17:07, 14 March 2017 by Rf (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search


Used for population genetics studies.


Quality trimming

Using trimmomatic, which allows various quality trimming options

java -jar $TRIMMOJARFILE PE -threads 6 -phred33 <path/to/forward_read> <path/to/reverse_read>  <forward_paired_name> <forward_unpaired_name><reverse_paired_name><reverse_paired_name> ILLUMINACLIP:/path/to/adapter/TruSeq3-PE-2.fa:2:30:10 LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:75


  • There are six arguments unattached to a particular option, consisting of the 2 inputs and the four outputs (because we will have paired and unpaired trimmed datasets). They are unattached because their exact position and order is assumed to be as above.

Map to mitochondrial reference genome


  • FQN1, the forward paired output from Trimmomatic
  • FQN2, the reverse paired output from Trimmomatic
  • MTSAM, the raw sam alignment to mito, unsorted
  • MTBAM, the alignment to mito, sorted and now in bam format.
  • UBAM, the unmapped bam reflecting exclusion of mitochondrial mappings

TFQ1=${SSD1}/${SDN}/${SSD1}_${SRN1}_${SLN1}_forward_paired.fq.gz # trimmed fq data file, forward paired reads TFQ2=${SSD2}/${SDN}/${SSD2}_${SRN2}_${SLN2}_reverse_paired.fq.gz # trimmed fq data file, reverse paired reads ACPTSZ=1000000 # acceptable size of a bam, a somewhat (only) risky way of seeing that it's OK.

REABAM=${SSD1}/${MOD}/aln-pe${SSD1}_${SRN1}_${SLN1}_sorted.bam REASAM=${SSD1}/${MOD}/aln-pe${SSD1}_${SRN1}_${SLN1}.sam

First we align to the mitochondrial reference:

bwa mem -t $NSLOTS $MITOREFLOC <quality_trimmed_forward_paired>  <quality_trimmed_reverse_paired> > <alignedtomitochondrial_samfile>

The resulting sam is sorted and converted to bam format:

samtools view -bSh <alignedtomitochondrial_samfile> | samtools sort - <alignedtomitochondrial_bamfile_withoutextension> -@ $NSLOTS

Note that the '.bam extension is not required in the output of samtools sort when its input and output are bam.

Re-generate readsets which will no longer map to mitochondrial genome

Actually, we're only interested in the reads that did not map to the mitochondrial, and we can extract them into the their own bam with:

samtools view -f4 -bh <alignedtomitochondrial_bamfile> > <unmappedotmito_bamfile> -@ $NSLOTS


  • -f4 specifies that only reads with a SAM FLAG setting of 1 on the third bit will be output, This is the setting that indicates the map did not not align. Essentially, this is the key part of this whole stage, it enables us obtain the reads which did not map to the mitichondrial reference, and also avoid those that did.

We can convert this extracted alignment back into FASTQ files with picard-tools and then compress as follows:

java -Xmx16g -jar $PICARDJARFILE SamToFastq I=${UBAM} F=${FQN1%.*} F2=${FQN2%.*}
pigz -f -p $NSLOTS ${FQN1%.*}
pigz -f -p $NSLOTS ${FQN2%.*}

These new reads can be aligned to the true reference

bwa mem -t $NSLOTS $REFLOC $FQN1 $FQN2 > $REASAM

The raw sam can be sorted and converted to bam to produce the alignment with the reads mapping to the mitochondrial reference excluded as follows:

samtools view -bSh ${REASAM} | samtools sort - ${REABAM%.*} -@ $NSLOTS

Marking Duplicates

As usual, the picard-tools function will be used for this. It can operate on multiple BAM files in one run, so we may apply it on a per-sample basis because we have several datasets for each sample in this pipeline.

java -Xmx32G -jar $PICARDJARFILE MarkDuplicates REMOVE_DUPLICATES=true I=<bamfile1> I=<bamfile2> I=... OPTICAL_DUPLICATE_PIXEL_DISTANCE=2500 O=<de-duplicated_bamfile> M=<marked_dup_metrics.txt>

Both the O (output) and M (metrics) options (filenames with relative directory paths) are chosen by the user.

We also generate an index for this bamfile which helps tools in downstream analysis.

samtools index <de-duplicated_bamfile> <de-duplicated_baifile>

Realigning after duplicate identification

An initial "target creator" step is required first because the IndelRealigner has a -targetIntervals option.

java -Xmx16G -jar /path/to/GenomeAnalysisTK.jar -T RealignerTargetCreator -R ${REFSEQLOC} -I <de-duplicated_bamfile> -o <chosen_targetcreator_intervals_filename>

Now the IndelRealigner tool can be invoked:

java -Xmx16G -jar /path/to/GenomeAnalysisTK.jar -T IndelRealigner -R ${REFSEQLOC} -targetIntervals <chosen_targetcreator_intervals_filename> -I <de-duplicated_bamfile> -o <chosen_realigned_bamfile>

An extra bit of quality filtering is also included here:

samtools view -bh -F4 -q 30 <chosen_realigned_bamfile> <realigned_qualityfiltered_bamfile> -@ $NSLOTS

Repeatmasking using a bed file

In this case, we have already run a repeatmasking exercise on the reference genome so we may safely apply it to the quality filtered bamfile produced in the previous step using bedtools:

intersectBed -abam <realigned_qualityfiltered_bamfile> -b <bedfile> -v | samtools view -h - | samtools view -bSh - > <no_repeats_bamfile>

where the bedfile represents the repeatmasked regions.