Skip to content

Latest commit

 

History

History
169 lines (116 loc) · 10.9 KB

SarahPaulIanDiscussion.md

File metadata and controls

169 lines (116 loc) · 10.9 KB

Discussion of pool seq papers to help us evaluate our own protocol

March 31st 2017

Kofler et al. 2016. Suitability of different mapping algorithms

  • no single one removed all artefacts.
  • novalign and bwa mem did pretty well for Fst.
  • Best to use two and use intersection of SNPs in VCF. The suggest bowtie2(g) and novalign(l) seemed to be the best combination.
  • There way of intersection is looking at the p-values for Fisher exact test at each site for each mapped. Use the p-value which is less significant. We (SM PK, ID) all think this may be too conservative as it could throw out real sites. Perhaps averaging values?
  • Is a better possibility to compare VCFs (VCFtools) for all alleles and average frequencies or something?
  • Worth noting that the free version of novalign is limited in how many cores can be used (grrr.).

April 5th 2017: Contrasting the differential computational pipelines from different research groups for pooled resequencing.

Petrov lab (ID)

I examined the protocols in two studies.

Bergland et al.. 2014. Genomic Evidence of Rapid and Stable Adaptive Oscillations over Seasonal Time Scales in Drosophila. PLoS Genetics.

The motivation was to examine how seasonal variation may result in "balanced polymorphism" (as there is no constant optimum), and may maintain SGV in populations.

They resequenced samples of D. melanogaster from populations sampled between 2003-2010. Each population sampled was between 50-200 individuals. Isofemales lines were made and progeny from this (around gen 5) were pooled to generate template DNA.

  1. Mapping to D.mel reference (v 5.39) using bwa (v 0.5.9-r16)
  • allowing for a maximum insert size of 800bp and no more than 10 mismatches/100bp. See below as they only used SNPs previously identified in the DGRP (so they were ok with higher overall error rates). i.e.
bwa aln -t10 -f r1.sai sample1.fastq
bwa aln -t10 -f r2.sai sample2.fasta
bwa sampe dmel_ref.fasta r1.fastq.trimmed.sai r2.fastq.trimmed.sai r1.fastq.trimmed r2.fastq.trimmed \
  | samtools view -Sb - > /yourdir/samples.bam

Files wre then indexed as normal samtools index ... and sorted as normal samtools sort ....

  1. PCR duplicates were removed using samtools (v0.1.18) with the bam files.
samtools rmdup ...
  1. local re-alignment around indels using GATK (v1.4-25). The new versions may not support this anymore (check manual). I think the target intervals were done once, and then used to perform re-alignment for each bam file.
java -Xmx8g -jar ~/GenomeAnalysisTK-1.4-25-g23e7f1b/GenomeAnalysisTK.jar -I \ 
  samples.sorted.dedup.bam -R dmel_ref.fasta -T RealignerTargetCreator -o \
  ./new_alignments/indelTargets/samples.intervals &
wait

java -Xmx8g -jar ~/GenomeAnalysisTK-1.4-25-g23e7f1b/GenomeAnalysisTK.jar -I \ 
  samples.sorted.dedup.bam -R dmel_ref.fasta -T IndelRealigner -targetIntervals \ 
  ./new_alignments/indelTargets/samples.intervals -o /dirs/m/samples.realign.bam & \
  wait
  1. Mapped SNPs and short indels using CRISP, excluding reads with base or mapping qualities below 10. The justification for using 10 (as opposed to 20 which is more common) is because they only used common SNPs that had previously been identified in DGRP.
  2. SNPs mapping to repetitive regions such as microsatellites and transposable elements, identified in the standard RepeatMasker library for D. melanogaster (obtained from http://genome.ucsc.edu) were excluded from analysis as were SNPs within 5 bp of polymorphic indels.
  3. SNPs with average minor allele frequency across all populations less than 15%, with minimum per-population coverage less than 10× or maximum per-population coverage greater than 400× were removed from analysis.
  4. Finally, to ensure that the examined SNPs were not artifacts of our pooled resequencing, we removed any SNP not present in the SNP tables provided by freeze 2 of the DGRP. Of the 1,500,000 SNPs initially identified, ∼500,000 SNPs remained after applying these filters.
  5. SNPs were annotated using SNPeff version 2.0.
  6. F_ST estimates were done as shown in the paper, but corrected for both the number of sampled chromosomes and number of reads at any site.

The second paper from the Petrov lab was: Machado et al.. Comparative population genomics of latitudinal variation in Drosophila simulans and Drosophila melanogaster. Mol. Ecol.

Motivation: Examine the influence of latitudinal variation and its impact with respect to selection and demography (using population genomics).

Some of the work was individual flies. For the pools, these used two samples. These were a few of the samples from the above study. They also used a correction for effective number of chromosomes (N_C) in the sample (see the paper).

  1. mapped raw reads (no trimming I guess) to D.mel genome v5.5 using bwa aln (v 0.7.9) and sampe with default parameters.
  2. Reads mapping to autosomes were down-sampled to match the N_C of the X chromosomes.
  3. Applied a Pool-seq error model.

I emailed the lead author (Heather Machado) and she confirmed that the SNP calls in this were those from the study above. She also mentioned that she is now using VarScan for SNP calling for pooled data.

Schlotterer lab (PK)

2 Papers looking at populations selected under hot and cold environmental fluctuations:

  1. Tobler et al.. 2014. Massive Habitat-Specific Genomic Response in D. melanogaster Populations during Experimental Evolution in Hot and ColdEnvironments
  2. Orozco-terWengel et al.. 2012. Adaptation of Drosophila to a novel labratory environment reveals temporally heterogeneous trajectories of selected alleles

Motivation:

  1. Identify which loci identified as selected sites (candidate loci) are real, and not false positives
  2. Looking at time series data for insight on genomic trajectories due to selection

Populations:

113 isofemale lines (orignally from Portugal, acclimated to lab for 5 generations) mixed to create base (F0) population (5 females per line == 565 founder population). Base populaton split into 10 populations, with 5 subjected to each fluctuating temperature environments: 5with hot environment (18-28 degrees) and 5 with cold environment (10-20 degrees). Pooled seqeuncing of 500 individuals for 3 repicates of the base population (B) and evolved populations at generation 15 (M) and generation 27 (E; used in 2. only) (Coverge range from 30-90 fold).

Pipeline:

A. Trimming: each refers to Popoolation for trimming, using a modified Mott algorithim from Phred (custom Perl script; trim.fastq.pl)

B. Mapping: BWA (1. version 0.5.8c, 2. version 0.5.7) to Drosophila referance 5.18 and Wolbachia

  1. Gave mapping parameters: -n 0.01 (error rate), -o 2 (gap opening), -d 12 & -e 12 (gap length), -l 150 (disable seed option)

C. Convert to SAM and remove unpaired map reads

  1. specified use of BWA sampe when a paired read could not mapped

D. Sam files filtered for quality > 20 with samtools

E. Convert to pileup format (samtools)

F. Repeatmasker (3.2.9) -- possibly same method, but explained different

  1. Use Repeatmasker to identify/mask indels and repeat sequences
  2. Repeatmasker to create a gff file to mask simple sequence repeats and transposable elements. Indels and the 5 flanking nucleotides (both sides)also masked if present in 1 populations and supported by 2 reads.

G. Converted to .sync file from which SNP's and allele frequencies derived

H. SNP filtering:

  1. Average across 3 replicates (weighted by coverage) - remove SNPs with coverage in top 2% of a given replicate and time point - min-count of 30 across all lab populations (extended data not in this study) which equated to ~ average of 2% minor allele frequency across all replicates and time points (could have lower frequencies for only middle time point in this study) - Approx. 1.45 million SNPs

  2. Following Criteria for SNPs - SNP occurance in atlease 2 replicate populations - minor allele covered by min 10 across all six populations analyzed (6??) - max coverage of 500 - exluded a section of 3R (1MB); a low freqeuncy haplotype absent in the base population but in evolved populations == likely caused by a single benefical mutation - exluded gene cluster on 3L due to higher coverage causing high false positive candidates

I. Identification of candidate SNPs using the CMH test

  1. Comparision of Hot-Base and Cold-Base with false discovery rate (FDR) of 0.01 - wright-fisher simulations of neutral evolution as estimate of drift

  2. Comparision of Allele Frequency Change (AFC; SNP-wise basis) between time points (B-M, M-E, & B-E) (replicates??) - wright-fisher simulations of neutral evolution as estimate of drift

J. Linkage effects: only within 1. Tobler et al.

  • neutral forward simulations of populations with 250 unique haplotypes for 15 generations to model recombination using MimicrEE
  • model correlated allele frequency dynamics caused by site linkage
  • Results: did not remove any candidates
  • look to Haploreconstruct (new method for dealing with haplotypes)

K. SNPeff: Only with 2. Orozco-terWengel et al.

  • SNPEFF 2.0.1 used to map all SNP's to genomic features
  • occurance within 200 bases of UTR (5' and 3') == possible regulatory motifs
  • meaure overrepresentation of selected SNPs in a given feature using chi-square tests

L. Gowinda Software

  • Permuation test, sampling all candidate SNPs mapped to a random position throughout genome (until # of random genes == number of candidate genes)
  • p-value is proportion of simulations with more genes (hypothesized to be selected for; in this case, thermal tolerance genes)

Pool lab (SM)

Bastide et al.. 2016. A Variable Genetic Architecture of Melanic Evolution in Drosophila melanogaster.

Point of study was to find the genetic basis of melanism.

Three dark pigemented populations were crossed with light color Zambian populations. Eight reciprocal crosses were set up for each of the combinations of crossed populations and the progeny were interbred until F20. F20 flies were scored for pigmentation and the most extreme 30 females were pooled together for sequencing for either dark or light color.

  • Library prep -> 100bp read lengths with 300bp inserts
  • Mapped to D.mel reference (v 5.57) using BWA (v 0.6.2-r126)
  • BAM files remapped using Stampy (v1.0.21)
  • Reads filtered for mapping quality of 20 and proper pairs with samtools (v0.1.18)
  • Picard (v1.109) use to remove PCR duplicates
  • Alignment around indels improved with GATK (v3.2)
  • Generated mpileups with PoPoolation2 (v1.201)
  • Calculated ancestry differentiation with their own perl scripts