Skip to content
/ brettEvolSSU1 Public

Scripts and workflows for Structural Variation in strains of Brettanomyces evolved for sulphur resistance

0 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

mroach-awri/brettEvolSSU1

BranchesTags

Repository files navigation

Bioinformatics scripts and workflows for publication "Adaptive evolution of sulfite tolerance in Brettanomyces bruxellensis"

Overview

Initial haploid assembly

Canu

canu -p 1499 -d CANU-NP-PB-2 \
    -genomeSize=24m \
    -stopOnReadQuality=false \
    -nanopore-raw 1499.fastq.gz \
    -nanopore-raw 1499.failed.fastq.gz \
    -pacbio-raw 1499.pb.fastq.gz \
    -minReadLength=3500 \
    -minOverlapLength=2500 \
    -corOutCoverage=100 \
    -correctedErrorRate=0.11

Purge Haplotigs

We originally tried deduplicating the contigs but haplotype switching made completely deduplicating the assembly impossible. We therefore mapped nanpore reads to the assembly unitigs, allowing complete deduplication which was necessary for only phasing the B haplome from the A1 + A2 haplomes.

minimap2 -t 12 1499.unitigs.fasta 1499.fastq.gz | samtools sort -@ 4 -m 1G -o aligned.bam
purge_haplotigs hist -g 1499.unitigs.fasta -b aligned.bam -t 16
purge_haplotigs cov -l 5 -m 95 -h 190
purge_haplotigs purge -g 1499.unitigs.fasta -c coverage_stats.csv
purge_haplotigs clip -p curated.fasta -h curated.haplotigs.fasta
purge_haplotigs place -p clip.fasta -h clip.haplotigs.fasta -f

Longshot

Longshot had a propensity to phase the divergent haplotype (B) while keeping the two more similar haplotypes (A1 and A2) together. This observation shaped the assembly strategy. Several contigs still had hapltype switching which needed to be split to properly bin the reads. Contigs > 50kb were manually inspected and sorted. contigs < ~50kb with roughly 2:1 size ratio were sorted into their anticipated bins (double size = A1 + A2, half size = B).

minimap2 -t 12 clip.FALC.fasta 1499.fastq.gz | samtools sort -@ 4 -m 1G -o clip.FALC.bam
# This script just runs Longshot contig-by-contig over multiple threads
perl multiLongShot.pl  clip.FALC.fasta.fai  clip.FALC.bam  16

Manually fix haplotype switches

Visualise contigs in IGV and identify haplotype switching. This is a list of haplotype switches. The Longshot-phased haplotypes are labelled either side of breakpoints to denote which of the haplotypes is the A1+A2 haplome. Contigs were split at these breakpoints and the BAM files were manually sorted into the separate directories for each haplome for read-binning.

000013F [1] 130k [2]
000006F [2] 124k [1]
000005F [1] 73.5k [2]
000009F [1] 82k [2] 127k [1]
000004F [1] 55k [2]
000003F [1] 31k [2] 103k [1] 129k
000014F [2] 82k [1]
000000F [1] 49k [2] 130k [1]
000011F [1] 64k [2]
000010F [2] 93k [1] 134k [2]
000024F [2] 15k [1]
000025F [2] 52k [1]
000027F [1] 74k [2]
000026F [2] 44k [1] 86k [2]
000029F [2] 70k [1]
000031F [1] 59k [2] 69k [1]
000036F [1] 59k [2] 69k [1]
000041F [1] 5k [2] 10k [1]
000045F [1] 80k [2]
000046F [1] 8k [2] 27k [1]
000053F [1] 66k [2]
000055F [2] 3k [1]
000057F [1] 28k [2]
000059F [1] 56k [2]
000060F [1] 43k [2]
000064F [1] 58k [2]
000068F [1] 36k [2] 45k [1]
000074F [1] 46k [2]
000085k [1] 30k [2]
000167F [2] 10k [1] 27k [2]

Read binning

Note: 'haploid' = B haplome directory and 'diploid' = A1A2 haplomes directory

# pull ids for each bin
for i in `ls haploid`; do 
    samtools view haploid/$i | awk '{print $1}'; 
done | gzip - > haploid.reads.gz

for i in `ls diploid`; do 
    samtools view diploid/$i | awk '{print $1}'; 
done | gzip - > diploid.reads.gz

# haploid read bin
perl getBinnedReads.pl 1499.fastq.gz  haploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.B.fastq.gz
perl getBinnedReads.pl 1499.fail.fastq.gz  haploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.B.fastq.gz
perl getBinnedReads.pl 1499.pb.fastq.gz  haploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.B.fastq.gz

# diploid read bin
perl getBinnedReads.pl 1499.fastq.gz  diploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.A1A2.fastq.gz
perl getBinnedReads.pl 1499.fail.fastq.gz  diploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.A1A2.fastq.gz
perl getBinnedReads.pl 1499.pb.fastq.gz  diploid.reads.gz  unassigned.reads.gz \
    | gzip - > 1499.A1A2.fastq.gz

Reassembly

# haploid canu assembly
canu -p 1499.B -d bin-B \
    -genomeSize=13m \
    -nanopore-raw 1499.B.fastq.gz \
    -nanopore-raw 1499.B.fail.fastq.gz \
    -pacbio-raw 1499.B.pb.fastq.gz \
    -stopOnReadQuality=false \
    -minReadLength=2500 \
    -minOverlapLength=2000 \
    -correctedErrorRate=0.11

# diploid canu assembly
canu -p 1499.B -d bin-A1A2 \
    -genomeSize=13m \
    -nanopore-raw 1499.A1A2.fastq.gz \
    -nanopore-raw 1499.A1A2.fail.fastq.gz \
    -pacbio-raw 1499.A1A2.pb.fastq.gz \
    -stopOnReadQuality=false \
    -minReadLength=2500 \
    -minOverlapLength=2000 \
    -correctedErrorRate=0.11

Purge Haplotigs was run on the read-binned assemblies following the same protocol as above. Both assemblies were concatenated to 1499.diploid.fasta with the prefixes 'A1A2_' or 'B_' prepended to contig names to denote haplomes.

Polishing

Racon

# map reads
minimap2 -t 12 1499.diploid.fasta ../../1499.fastq.gz -x map-ont \
    | gzip - > NP.paf.gz

# run racon
racon -t 16 ../../1499.fastq.gz NP.paf.gz 1499.diploid.fasta \
    > 1499.racon.fasta

Pilon

# index fasta
bwa index 1499.racon.fasta

# map reads
bwa mem -t 12 1499.racon.fasta \
    <( zcat ../../1499_S4_L001_R1_001.fastq.gz ../../1499_S29_L001_R1_001.fastq.gz) \
    <(zcat ../../1499_S4_L001_R2_001.fastq.gz ../../1499_S29_L001_R2_001.fastq.gz) \
    | samtools sort -@ 4 -m 1G -o pe.bam

# run pilon
java -jar ~/pilon.jar --genome 1499.racon.fasta --frags pe.bam --threads 16

# rename file
mv pilon.fasta 1499.racon.pilon.fasta

# trim '_pilon' from contig names
cat 1499.racon.pilon.fasta | sed 's/_pilon//' > 1499.diploid.newPolished.fasta

Annotate

We annotated genes with Augustus and used these for initially investigating the key genes, but the results in the manuscript relating to genes affected by SVs all relate to the AWRI2804 reference annotations.

augustus --species=saccharomyces_cerevisiae_S288C 1499.diploid.newPolished.fasta > 1499.gff

# extract prot seqs
cat 1499.gff | perl -e \
    '$/ = "end gene"; 
    while(<>){
        my$gen;
        if(m/start gene (\w+)/){
            $gen=$1;
        }else{
            die;
        }
        s/#//g;
        if(m/protein sequence = \[([\w\s]+)\]/){
            $s=$1;
            $s=~s/\s//g;
            print ">$gen\n$s\n";
        }
    }' \
    > 1499.faa

# prot names
cat 1499.faa \
    | parallel -j 8 --recstart '>' --blocksize 10k --pipe \
        blastp -query - -db /home/mike/blastdb/uniprot_sprot.fasta \
        -evalue 1e-5 -num_alignments 1 -outfmt 6 \
        | gzip - > 1499.uniprotkb.outfmt6.gz

# add the draft names as descriptions
perl addDraftNames.pl 1499.diploid.uniprotkb.outfmt6.gz uniprot_sprot.names.tsv 1499.faa > tmp
mv tmp 1499.faa

SV calling

Sniffles

Most of the Sniffles SV calls overlapped with the below SV calling pipeline, so they were left out of the final manuscript.

# mapping
ngmlr -x ont -t 8 -r 1499.diploid.fasta -q ../1499.fastq.gz | samtools sort -@ 4 -m 1G -o 1499.bam -T tmp
ngmlr -x ont -t 8 -r 1499.diploid.fasta -q ../1499_11A.fastq.gz | samtools sort -@ 4 -m 1G -o 1499_11A.bam -T tmp
ngmlr -x ont -t 8 -r 1499.diploid.fasta -q ../1499_17A.fastq.gz | samtools sort -@ 4 -m 1G -o 1499_17A.bam -T tmp
ngmlr -x ont -t 8 -r 1499.diploid.fasta -q ../1499_52C.fastq.gz | samtools sort -@ 4 -m 1G -o 1499_52C.bam -T tmp

# call SVs with sniffles
for i in 1499 1499_11A 1499_17A 1499_52C; do
    sniffles -m $i.bam -b $i.bedpe -t 8
    sniffles -m $i.bam -v $i.vcf -t 8
    done

# quick filter
for i in 1499 1499_11A 1499_17A 1499_52C; do
    head -1 $i.bedpe > $i.filt.bedpe
    cat $i.bedpe | awk '$12>30{print}' >> $i.filt.bedpe
    done

# manually curate with genome-ribbon: doi.org/10.1101/082123

Read-depth and SNP-based SV pipeline

The pipeline uses SNP and read-depth comparisons to annotate SV features (deletions, duplications, conversions) in theevolved isolates. Annotations are manually inspected and adjusted and the pipeline is rerun to plot a karyogram of the SVs relative to the reference assembly for AWRI2804.

# align 'diploid' queries to haploid reference
nucmer -b 500 -c 40 -d 0.5 -g 200 -l 12 -t 16 \
    AWRI2804_scaffold.fasta 1499.diploid.fasta -p 2804.1499
nucmer -b 500 -c 40 -d 0.5 -g 200 -l 12 -t 16 \
    AWRI2804_scaffold.fasta 1613.H1.fasta -p 2804.1613

# filter all query to best ref
delta-filter -q 2804.1499.delta > 2804.1499.qdelta
delta-filter -q 2804.1613.delta > 2804.1613.qdelta

# create a transformation table
show-coords -HTl 2804.1499.qdelta \
    | awk '$5>1000{if($4<$3){print $10"\t"$1"\t"$2"\t"$11"\t"$4"\t"$3"\t-\t"$9}else{print $10"\t"$1"\t"$2"\t"$11"\t"$3"\t"$4"\t+\t"$9}}' \
    > 1499.to.2804.trn
show-coords -HTl 2804.1613.qdelta \
    | awk '$5>1000{if($4<$3){print $10"\t"$1"\t"$2"\t"$11"\t"$4"\t"$3"\t-\t"$9}else{print $10"\t"$1"\t"$2"\t"$11"\t"$3"\t"$4"\t+\t"$9}}' \
    > 1613.to.2804.trn

# make directories
snakemake  -s findGenomeModifications.py

# manually copy the sequencing data and translation tables 

# find draft conversions
snakemake  -s findGenomeModifications.py  -j 16

# manually curate the draft SV calls in IGV

# produce final plots
snakemake  -s findGenomeModifications.py  -j 16

Clipped Read Island-based SV calling for 2804

AWRI2804, a haploid strain, didn't have any notable changes identified from the read-depth-based SV calling, and some apparent transpositions were not being called by Sniffles. We therefore attempted to identify transpositions based on concentrations of clipped long-read mapping, comparing to the control strian.

# extract soft-clipped reads from bam file
samtools view -h ../BAMS/2804_50G_control_23B.2804.bam \
    | perl -e 'while(<>){@l=split/\s+/;if(($l[0]=~/^@/) or ($l[5]=~/\d\d\d\dS/)){print $_;}}' \
    | samtools sort - > 23B.clipped.bam

# manually inspect in IGV and copy coordinates of clipped read islants to file:
# clippedReadIslands.bed; this wouldn't be practical for large genomes.

# get the clipped read IDs of soft-clipped reads in clipped island regions
grep 23B clippedReadIslands.bed > 23B.clipped.bed
bedtools intersect -a 23B.clipped.bam -b 23B.clipped.bed \
    | samtools view \
    | awk '{print $1}' > 23B.clipped.readIDs 

# get the reads themselves
zcat ../SEQ/upload/2804_50G_23B.fastq.gz \
    | ./fastqByReadID.pl 23B.clipped.readIDs \
    | gzip - > 23B.clippedReads.fastq.gz 

# reassemble these reads
minimap2 -x ava-ont -t 12 23B.clippedReads.fastq.gz 23B.clippedReads.fastq.gz \
    | gzip - > 23B.paf
miniasm -f 23B.clippedReads.fastq.gz 23B.paf > 23B.gfa
awk '$1 ~/S/ {print ">"$2"\n"$3}' 23B.gfa > 23B.clipAsm.fasta

# quick polish
minimap2 -t 12 23B.clipAsm.fasta 23B.clippedReads.fastq.gz > 23B.asm.paf
racon 23B.clippedReads.fastq.gz 23B.asm.paf 23B.clipAsm.fasta -t 12 > 23B.asm.racon.fasta

About

Scripts and workflows for Structural Variation in strains of Brettanomyces evolved for sulphur resistance

Resources

Readme
Activity

Stars

0 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases 1

1.0 Latest
Dec 21, 2020

Packages

No packages published

Languages