/posseleff_empirical

Nextflow pipeline for analyzing empirical WGS data for the effect of positive selection on IBD-based inference

Primary LanguageNextflowMIT LicenseMIT

poseleff_empirical

The code in this repository analyzes empirical WGS (Whole Genome Sequencing) data to assess the effect of positive selection on IBD (Identity by Descent)-based inference. This analysis process encompasses variant and sample filtering, multiplicity of infection inference, deconvolution and imputation (for both monoclonal and a subset of polyclonal samples), IBD calling, and IBD-based inferences of effective population size (Ne) and population structure. These functions are detailed and implemented across five modules, as described in the "Pipeline outline" section below.

System requirement and Software evironment

The pipeline has been tested on Linux Operation system and can be easily adapted to MacOS with simple changes. Software dependencies and the version numbers are specified in the './env.yaml' Conda recipe. Additional depencies that are not available from Conda are specified in the installation instruction below. The overall installation time is about 5-30 minutes. The process is relatively long due to the compliation of the R package 'moimix' and its dependencies.

  1. Install Nextflow
  2. Install Conda
  3. Create a Conda evironment from a provided recipe: conda env create -f ./env.ymal
  4. Activate conda environment: conda activate empirical
  5. Install moimix package: Run R and in R console run:
install.packages("BiocManager")
BiocManager::install("bahlolab/moimix", build_vignettes = TRUE)
  1. Modify (recombination rate) and install hmmibd:
    git clone https://github.com/glipsnort/hmmIBD.git; cd hmmIBD
    # checkout a specific version for reproducibility
    git checkout a2f796ef8122d7f6b983ae9ac4c6fba35afcd3aa
    sed -i -e 's/const double rec_rate = 7.4e-7/const double rec_rate = 6.67e-7/' hmmIBD.c
    # use the complier from the `empirical` conda environtment
    x86_64-conda_cos6-linux-gnu-gcc -o hmmIBD -O3 -lm -Wall hmmIBD.c
    cd ..
    cp hmmIBD/hmmIBD bin/hmmIBD
    rm -rf hmmIBD

Pipeline outline

The entire pipeline is organized into five modules, which include:

  • 01_filter_vcf.nf: Filters VCF files by per-sample and per-site missingness and minor allele frequency.
    • This involves two rounds of filtering to help retain more samples/sites.
  • 02_calc_pop_fws.nf: Calculates Fws statistics using the moimix R package.
    • Data is organized into populations (Pfv6 classification).
    • Within each population, Fws is inferred for individual samples.
  • 03_phase_impute_monoclonal.nf:
    • Filters samples and retains only monoclonal samples.
    • Deconvolutes each monoclonal sample (dEploid).
    • Imputes phased genotype data within each population (Beagle).
    • Merges single-sample genotypes back to multiple-sample, phased & imputed VCF files.
  • 04_phase_impute_polyclonal.nf:
    • Uses a subset of phased & imputed monoclonal samples to build a deconvolution panel for polyclonal samples within each population.
    • Filters samples and retains only polyclonal samples.
    • Deconvolutes genotype data for each polyclonal sample using a population-specific panel (dEploid-IBD).
    • Filters polyclonal samples, retaining only those with a dominant clone/genome.
    • Combines phased polyclonal samples with the above monoclonal samples.
    • Imputes the combined genotype data (monoclonal/polyclonal samples) within each population (Beagle).
  • 05_ibd_ne_ifm.nf main.nf:
    • Calls IBD using hmmIBD.
    • Processes IBD to prepare input (with or without selection correction) for IBDNe analysis.
    • Processes IBD to prepare input (with or without selection correction) for Infomap analysis.
    • Runs IBDNE to infer effective population size.
    • Runs Infomap to infer population structure.

The five modules will run sequentially if using main.nf as the the entry point.

General notes:

  • Thresholds can be configured in the top lines of each module file.
  • The pipeline is designed to run in parallel for large datasets.
  • Each module is independent of the others, provided that valid inputs are supplied.

How to use the pipeline

After install the pipeline, first prepare the input files and then run the nextflow pipeline with the input files.

Input

1. Vcf file table:

The VCF file table is organized in a two-column format:

  • Header: None
  • Delimiter: Tab
  • Column 1: Represents the genome interval
    • Format for genome interval: [Chromosome]:[Start]-[End]
  • Column 2: Provides the full path to the joint-call VCF file corresponding to the specified genome interval
    • It's essential that VCF files for different genome intervals include the same set of samples.
    • The VCF file should contain annotations for the hard filter/vqsr filter within the FILTER columns. Refer to the output from the snp_call_nf pipeline for more information.
    • For those interested in reproducing the VCF files via snp_call_nf pipeline, you can find the accession numbers of WGS data for the samples used in this study within the samples folder.

Example file interval_vcf_table.tsv

Pf3D7_01_v3:1-459121	    /full/path/to/filePf3D7_01_v3:1-459121.vqsrfilt.SNP.vcf.gz
Pf3D7_01_v3:459122-640851   /full/path/to/filePf3D7_01_v3:459122-640851.vqsrfilt.SNP.vcf.gz
Pf3D7_02_v3:1-448875	    /full/path/to/filePf3D7_02_v3:1-448875.vqsrfilt.SNP.vcf.gz
Pf3D7_02_v3:448876-947102   /full/path/to/filePf3D7_02_v3:448876-947102.vqsrfilt.SNP.vcf.gz
...
Pf3D7_14_v3:2121871-2706733 /full/path/to/filePf3D7_14_v3:2121871-2706733.vqsrfilt.SNP.vcf.gz
Pf3D7_14_v3:2706734-3291936 /full/path/to/filePf3D7_14_v3:2706734-3291936.vqsrfilt.SNP.vcf.gz
Pf3D7_14_v3:568025-1073306  /full/path/to/filePf3D7_14_v3:568025-1073306.vqsrfilt.SNP.vcf.tz

If the VCF file is not in 'gz' format, use bgzip xx.vcf to convert it to *.vcf.gz format.

2. Population table

The population table is a two-column table

  • Header: none
  • Delimiter: tab
  • The first column contains the population name
  • The second column specifies the path to a file that contains all samples from the given population
    • In the list file, each sample name appears on a separate line
    • The sample names should be the same as in the input VCF files
    • All samples in the VCF files should be contained in one of the population sample list files

Example file subpops_table.tsv

CAF     subpops/CAF.txt
EAF	    subpops/EAF.txt
ESEA    subpops/ESEA.txt
Lab	    subpops/Lab.txt
OCE	    subpops/OCE.txt
SAM	    subpops/SAM.txt
SAS	    subpops/SAS.txt
WAF	    subpops/WAF.txt
WSEA    subpops/WSEA.txt

Example file subpops/ESEA.txt

Sample1
Sample2
Sample3
...
Samplen

The same format is shared maong all populaton sample files such as subpops/EAF.txt (East Africa), subpops/ESEA.txt (Eastern Southeast Asia), subpops/Lab.txt (Laboratory Strains), subpops/OCE.txt (Oceania), subpops/SAM.txt (South America), subpops/SAS.txt (South Asia), subpops/WAF.txt (West Africa), and subpops/WSEA.txt (Western Southeast Asia).

Run the pipeline

conda activate empirical
nextflow ./main.nf -profile -resume -profile sge \
    --input_vcf_fn ./interval_vcf_table.tsv \
    --input_subpop_fn ./subpops_table.tsv \
    --outdir results

For large datasets, using a cluster such as SGE is recommended. An example sge profile is provided in the nextflow.config file and should be adjusted to fit your cluster system.

If run on a local computer, please remove the -profile sge option from the above command.

Test data

Test data for module 1-4 are not directly provided as they are extremely large but can be easily generated by running the ena_test test in the [snp_call_nf pipeline'] (https://github.com/bguo068/snp_call_nf).

A small dataset is provided, however, for module 5. The test of module 5 can be run with the following command:

nextflow 05_ibd_ne_ifm.nf -profile sge --vcf 'test_data/*.vcf.gz'

Check the Output

  • The results/01_filt_vcf/08_maf_bcf folder contains the final result files for module 1, i.e., VCF files filtered for samples or sites with high missingness and sites with low minor allele frequency.
  • The results/02_pop_fws/04_moimix_fws folder contains the final results for module 2, i.e., a table of Fws statistics for each sample. Each population will have a separate table.
  • The results/04_phase_impute_polyclonal/05_phased_imputed_polyclonal_monoclonal folder contains the final result files for modules 3 and 4, i.e., combined VCF files that have been deconvoluted and imputed. Each population will have a separate VCF file.
  • The results/05_ibdanalysis/02_callibd/hmmibd folder contains the IBD results obtained using hmmIBD.
  • The results/05_ibdanalysis/{population}/ne_output folder contains the results of IBDNe (effective population size).
  • The results/05_ibdanalysis/{population}/ifm_output folder contains the results of Infomap (population structure).

Citations

If you find this repository useful, please cite our preprint:

Guo, B., Borda, V., Laboulaye, R., Spring, M. D., Wojnarski, M., Vesely, B. A., Silva, J. C., Waters, N. C., O'Connor, T. D., & Takala-Harrison, S. (2023). Strong Positive Selection Biases Identity-By-Descent-Based Inferences of Recent Demography and Population Structure in Plasmodium falciparum. bioRxiv : the preprint server for biology, 2023.07.14.549114. https://doi.org/10.1101/2023.07.14.549114

Other citations:

  • IBDNe

Browning, S. R., & Browning, B. L. (2015). Accurate Non-parametric Estimation of Recent Effective Population Size from Segments of Identity by Descent. American journal of human genetics, 97(3), 404–418. https://doi.org/10.1016/j.ajhg.2015.07.012

  • Infomap algorithm

Rosvall, M., & Bergstrom, C. T. (2008). Maps of random walks on complex networks reveal community structure. Proceedings of the National Academy of Sciences of the United States of America, 105(4), 1118–1123. https://doi.org/10.1073/pnas.0706851105

  • dEploid:

Zhu, S. J., Almagro-Garcia, J., & McVean, G. (2018). Deconvolution of multiple infections in Plasmodium falciparum from high throughput sequencing data. Bioinformatics (Oxford, England), 34(1), 9–15. https://doi.org/10.1093/bioinformatics/btx530

  • dEploid-IBD:

Zhu SJ, Hendry JA, Almagro-Garcia J, Pearson RD, Amato R, Miles A, Weiss DJ, Lucas TC, Nguyen M, Gething PW, Kwiatkowski D, McVean G; Pf3k Project. The origins and relatedness structure of mixed infections vary with local prevalence of P. falciparum malaria. Elife. 2019 Jul 12;8:e40845. doi: 10.7554/eLife.40845. PMID: 31298657; PMCID: PMC6684230.

  • hmmIBD

Schaffner SF, Taylor AR, Wong W, Wirth DF, Neafsey DE. hmmIBD: software to infer pairwise identity by descent between haploid genotypes. Malar J. 2018 May 15;17(1):196. doi: 10.1186/s12936-018-2349-7. PMID: 29764422; PMCID: PMC5952413.

  • Beagle

Browning, B. L., Zhou, Y., & Browning, S. R. (2018). A One-Penny Imputed Genome from Next-Generation Reference Panels. American journal of human genetics, 103(3), 338–348. https://doi.org/10.1016/j.ajhg.2018.07.015

  • scikit-allel (calculate iHS statistics)

https://zenodo.org/records/8326460

https://github.com/cggh/scikit-allel

Related Private Repository:

For internal use only, refer to the following link: https://github.com/bguo068/posseleff_empirical_arxiv.