Skip to content

SARS-CoV-2 lineage importations and spread are reduced after nonpharmaceutical interventions in phylogeographic analyses

Notifications You must be signed in to change notification settings

hzi-bifo/covid-germany-mcmc

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

26 Commits
 
 
 
 
 
 

Repository files navigation

SARS-CoV-2 lineage importations and spread are reduced after nonpharmaceutical interventions in phylogeographic analyses

Goliaei S. (1,2), Foroughmand-Araabi M.H. (1,2), Roddy A. (1,2), Weber A. (3), Översti S. (3), Kühnert D. (3,4) McHardy A.C. (1,2,+)

  • 1 Computational Biology of Infection Research, Helmholtz Centre for Infection Research, Braunschweig, Germany,
  • 2 Braunschweig Integrated Centre of Systems Biology (BRICS), Technische Universität Braunschweig, Braunschweig, Germany
  • 3 Transmission, Infection, Diversification and Evolution Group, Max-Planck Institute of Geoanthropology, Jena, Germany
  • 4 German COVID Omics Initiative (deCOI), Bonn, Germany
  • + Corresponding author

The computational workflow for inference of importation lineages of SARS-Cov-2 into Germany. The workflow was adapted from [https://github.com/COG-UK/UK-lineage-dynamics-analysis], for processing a 20-fold larger dataset (original workflow for analysis of first UK wave processed 50K samples and this analysis was handling 1.8M sequence samples).

Note that because of the GISAID terms of use genomic sequences cannot be shared in this (public) repository. Instead, in the metadata files (all the tsv files under the analyses/*/results folders), we kept only the ID of sequences from GISAID data. The GISAID Accession ID for the samples after subsamping and filterings are available in files results of in/out ratio subsampling with mash identicals, results of in/out ratio subsampling, results of in/out ratio subsampling with identicals, results of fixed bucket size subsampling with mash identicals, results of fixed bucket size subsampling, results of fixed bucket size subsampling with identicals, results of case-ratio subsampling with mash identicals, results of case-ratio subsampling, results of case-ratio subsampling with identicals.

The preprint of the manuscript is available at medRxiv.

Workflow overview:

File analyses/phylogenetic/run-2.sh is the script containing all the steps. The workflow includes some optional steps, some steps for testing the results. Some steps are very time-consuming, so one may want to skip those steps.

Setting up the environment:

First, we create a new folder and inside it clone this github repository.

mkdir covid-mcmc
cd covid-mcmc
git clone git@github.com:hzi-bifo/covid-germany-mcmc.git

Required datasets.

We assume that following data are located at:

  • Raw sequences at data/data/gisaid-20210602-raw.fa.
  • Metadata file at data/data/gisaid-20210602-metadata.tsv.
  • (Optional) Sequence aligment file at data/data/mmsa_20210622_masked.fa Note that, these files should be downloaded from the GISAID website after signing up and accepting the agreement form, thus these files are not provided in the repository. The files could be find in the GISAID website following these paths: Raw sequences from GISAID -> EpiCov -> Downloads -> FASTA, metadata file from GISAID -> EpiCov -> Downloads -> metadata, sequence alignment file from GISAID -> EpiCov -> Downloads -> MSA masked.
mkdir -p data/data/

Download previous files and put them in the mentioned locations.

Required applications

Install the following applications:

  • Thorney BEAST.
  • The Thorney BEAST package contains the beast application.
  • The Thorney BEAST package also contains the treeannotator application.
  • (Optional) mash application, for reproducing results with added identical sequences. Create a conda environment with the following libraries:
  • boost and boostcpp Note: Environment variable PATH should contain above mentioned installed applications.
mkdir bin/
cd bin
wget https://github.com/beast-dev/beast-mcmc/releases/download/v1.10.5pre_thorney_v0.1.1/BEASTv1.10.5pre_thorney_0.1.1.zip
unzip -q BEASTv1.10.5pre_thorney_0.1.1.zip
PATH=$PWD/BEASTv1.10.5pre_thorney_0.1.1/bin:$PATH
cd ../
# You can check if beast works fine by running the following command
beast -version

# Creating a conda environment and installing packages
conda create --name covid-uk
conda activate covid-uk
conda install boost-cpp boost -c conda-forge

Building executables from the current repo.

Applications from the current repository should be compiled and executable files produces. Code of some applications are not located in the current repository. So clone the repository https://github.com/hzi-bifo/covid-germany-mcmc-phylogeo-tools and put it at ../../../phylogeo-tools (relative to the working directory). Then you can check if the symlinked files are pointing to the current source codes (e.g. scripts/state.h file relative to the working directory).

# It should be executed inside the covid-mcmc folder, we created first.
git clone git@github.com:hzi-bifo/covid-germany-mcmc-phylogeo-tools.git phylogeo-tools
git clone git@github.com:hzi-bifo/covid-germany-mcmc-one-tree-public.git one-tree

To create executables, on the folder scripts/ it is enough to run following command:

cd covid-germany-mcmc/analyses/phylogenetic/scripts/
make
cd ../../../../

Note that the makefile uses environment variable CONDA_PREFIX based on which include directory $(CONDA_PREFIX)/include and lib directory $(CONDA_PREFIX)/lib are addressed.

Setting environment variables:

To start the executions, we first change directory to the working folder as

cd huge-lineage-dynamics/analyses/phylogenetic/
  • Environment variable LD_LIBRARY_PATH should point to a place with libboost_programoptions.
  • All the executions could be done from working directory analyses/phylogenetic/ relative to the home of the github repository.
  • Following variables are reporesenting current data timestamps. So, please keep it as so.
DATE_TREE=20210602
DATE_METADATA=20210602
DATE_SEQ=20210602
  • State should be "Germany". For other countries, some modifications should be made.
STATE=Germany
  • TWD points to a place with temporary but large free space.
TWD=/tmp
  • SUB_TREES are the important subtrees found after subsampling.
SUB_TREES="A B.1.1.7 B.1.1.519 B.1.1.70 B.1.1.317 B.1.177 B.1.160 B.1.221 B.1.36 B.1.258 B.1.351 C"

Cleaning:

  • Input (not in repository): ../../../data/data/gisaid-$DATE_SEQ-raw.fa
  • Output (not in repository): ../../../data/data/gisaid-$DATE_SEQ-Wu1.fa $TWD/gisaid-$DATE_SEQ/gisaid-$DATE_SEQ-raw.fa ../../../data/data/gisaid-$DATE_SEQ.fa Note that the file ../../../data/data/gisaid-$DATE_SEQ-raw.fa should be edited to contain only sequence hCoV-19/Wuhan/WH04/2020 after execution of the first step.
#cleaning
grep -A1000 -F 'hCoV-19/Wuhan/WH04/2020' ../../../data/data/gisaid-$DATE_SEQ-raw.fa > ../../../data/data/gisaid-$DATE_SEQ-Wu1.fa
#edit then ..., label: hCoV-19/Wuhan/WH04/2020|EPI_ISL_406801|2020-01-05
DIR=`pwd`
mkdir $TWD/gisaid-$DATE_SEQ/
cd $TWD/gisaid-$DATE_SEQ/
ln -s $DIR/../../../data/data/gisaid-$DATE_SEQ-raw.fa .
bash $DIR/sarscov2phylo/clean_gisaid.sh -i gisaid-$DATE_SEQ-raw.fa -o gisaid-$DATE_SEQ.fa -t 30
mv gisaid-$DATE_SEQ.fa $DIR/../../../data/data/

cd $DIR

Subsampling

  • Input: ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv ../../../data/data/gisaid-$DATE_SEQ.fa
  • Output: results/pangolin-$DATE_TREE-r0.tree results/gisaid-$DATE_TREE-metadata-sampled.tsv ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled0.fa ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled.fa results/gisaid-$DATE_TREE-?-samples0.txt (results/gisaid-20210602-B.1.1.7-samples0.txt, ...) (for different subtrees with their names substitutes of $SUB_TREES character '?') results/gisaid-$DATE_TREE-?-sampled0.tree (results/gisaid-$DATE_TREE-B.1.1.7-sampled0.tree, ...) (for different subtrees of $SUB_TREES with their names substitutes character '?')
## SUBSAMPLING: 
#V3 (CURRENT): build tree as pangolin tree: 
./scripts/tree-build-as-pangolin --out results/pangolin-$DATE_TREE-r0.tree --metadata ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv --seqs ../../../data/data/gisaid-$DATE_SEQ.fa  -l Germany nonGermany
# we use following for filtering non-good samaple (e.g. sample without complete date)
./scripts/phylogeo_sankoff_general --in results/pangolin-$DATE_TREE-r0.tree --out results/gisaid-$DATE_TREE-$STATE.tree --metadata ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv --location_label $STATE non$STATE --cond 2 "==" Germany

./scripts/sample-evenly --in results/gisaid-$DATE_TREE-$STATE.tree --out results/gisaid-$DATE_TREE-sampled.tree~ -l $STATE non$STATE --samples-out results/gisaid-$DATE_TREE-samples.txt~ --metadata ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv --bucket-size 100 25 --special 5 $SUB_TREES --keep 1000 10000 $SUB_TREES
cp results/gisaid-$DATE_TREE-sampled.tree~ results/gisaid-$DATE_TREE-sampled.tree
cp results/gisaid-$DATE_TREE-samples.txt~ results/gisaid-$DATE_TREE-samples.txt

# We can check the statistics of the trees and subtrees with following commands. The SUB_TREES variable is calculated beased on following statistics.
./scripts/tree-stat --in results/gisaid-$DATE_TREE.tree -l $STATE non$STATE --large 1000 --depth 10
./scripts/tree-stat --in results/gisaid-$DATE_TREE-sampled.tree -l $STATE non$STATE --large 500 --depth 30

./scripts/extract-metadata.R results/gisaid-$DATE_TREE-samples.txt ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv results/gisaid-$DATE_TREE-metadata-sampled.tsv
cut -f1 results/gisaid-$DATE_TREE-metadata-sampled.tsv | tail -n +2  > results/gisaid-$DATE_TREE-sample-names.txt
./scripts/alignment-filter ../../../data/data/gisaid-$DATE_SEQ.fa results/gisaid-$DATE_TREE-sample-names.txt > ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled0.fa
grep '>' ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled0.fa | sort | uniq -c | grep -v '^ *1 ' | sort -nr
./scripts/rename-alignment-ids results/gisaid-$DATE_TREE-metadata-sampled.tsv < ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled0.fa > ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled.fa

# Extracting data of the SUB_TREES
./scripts/partition-by-name --in results/gisaid-$DATE_TREE-sampled.tree --par $SUB_TREES --samples "results/gisaid-$DATE_TREE-?-samples0.txt" --trees "results/gisaid-$DATE_TREE-?-sampled0.tree" -l $STATE non$STATE --print-annotation false --print-internal-node-label false

Creating phylogeny for each sub-tree of SUB_TREES:

  • Input: ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled.fa results/gisaid-$DATE_TREE-$X-samples0.txt results/gisaid-$DATE_TREE-$X-seq0.fa ../../../data/data/gisaid-$DATE_SEQ-Wu1.fa
  • Output: results/gisaid-$DATE_TREE-$X-1.tree
# Creating sequences for each sub-tree
for X in $SUB_TREES; do
        ./scripts/alignment-filter ../../data/phylogenetic/gisaid-$DATE_SEQ-sampled.fa results/gisaid-$DATE_TREE-$X-samples0.txt > results/gisaid-$DATE_TREE-$X-seq0.fa
        cat ../../../data/data/gisaid-$DATE_SEQ-Wu1.fa >> results/gisaid-$DATE_TREE-$X-seq0.fa
done

Following commands creates the phylogenies via [sarscov2phylo https://github.com/roblanf/sarscov2phylo] pipeline. Execution of following commands are time consuming.

for X in $SUB_TREES; do
        DIR=`pwd` ;
        rm -rf results/fasttree/$DATE_TREE-$X/ ;
        mkdir -p results/fasttree/$DATE_TREE-$X/ ;
        ln -s $DIR/results/gisaid-$DATE_TREE-$X-seq0.fa results/fasttree/$DATE_TREE-$X/gisaid-$DATE_TREE-$X-seq0.fa ;
        cp scripts/qrun-fasttree.sh results/fasttree/$DATE_TREE-$X/qrun.sh ;
        cd results/fasttree/$DATE_TREE-$X/ ;
        # Following command was made for submitting a job. To be checked in this version
        bash qrun.sh $DATE_SEQ $DATE_TREE $X ;
        cd - ;
done

MCMC for phylogeny topology reconstruction:

  • Input: results/fasttree/$DATE_TREE-$X/ft_FBP.tree results/gisaid-$DATE_TREE-$X-samples0.txt results/gisaid-$DATE_TREE-metadata-sampled.tsv data/X.fixedRootPrior.skygrid-template-thorney.xml
  • Output: results/beast/run/$X-$i/ for $X in $SUB_TREES and $i in 31-35, results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.log
  • Intermediate files: results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.xml

Declaring initial clock rate for MCMCs.

declare -A SUB_TREE_CLOCK_RATE=(
                [B.1.1.7]=0.00075
                [B.1.1.317]=0.00075
                [B.1.1.214]=0.00075
                [B.1.160]=0.00075
                [B.1.221]=0.00075
                [B.1.36]=0.00075
                [B.1.351]=0.00075
                [P]=0.00075
                [A]=0.00075
                [C]=0.00075
                [B.1.617.2]=0.00075
                [B.1.1.519]=0.00075
                [B.1.258]=0.00075
                [B.1.1.70]=0.00075
                [B.1.177]=0.00075
)

Creating XMLs for BEAST.

for X in $SUB_TREES; do
        cp results/fasttree/$DATE_TREE-$X/ft_FBP.tree results/gisaid-$DATE_TREE-$X-1.tree ;
        ./scripts/extract-metadata.R results/gisaid-$DATE_TREE-$X-samples0.txt results/gisaid-$DATE_TREE-metadata-sampled.tsv results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv

        # B.1.1.70
        grep -v EPI_ISL_2333117 results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv > x; cat x > results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv
        grep -v EPI_ISL_2333526 results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv > x; cat x > results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv

        #B.1.1.177
        grep -v EPI_ISL_2333528 results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv > x; cat x > results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv


        # B.1.258
        grep -v EPI_ISL_2333525 results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv > x; cat x > results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv

        # B.1.1.7
        grep -v EPI_ISL_2357883 results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv > x; cat x > results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv

        ./scripts/phylogeo_sankoff_general --in results/gisaid-$DATE_TREE-$X-1.tree --out results/gisaid-$DATE_TREE-$X-2.tree --metadata results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv --location_label $STATE non$STATE --cond 2 "==" Germany  --print-annotation false --print-internal-node-label false --single-child false --samples results/gisaid-$DATE_TREE-$X-samples1.txt


        ./scripts/contract_short_branch --in results/gisaid-$DATE_TREE-$X-2.tree --out results/gisaid-$DATE_TREE-$X-cont.tree --short 5e-6 --location_label $STATE non$STATE --print-annotation false --print-internal-node-label false --contract_leaf_enabled false

        ./scripts/fill-template.py --template data/X.fixedRootPrior.skygrid-template-thorney.xml --name $X --sample results/gisaid-$DATE_TREE-$X-samples1.txt --tree results/gisaid-$DATE_TREE-$X-cont.tree --tree_init results/gisaid-$DATE_TREE-$X-2.tree --metadata results/gisaid-$DATE_TREE-metadata-sampled.tsv --out results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.xml --loc Germany --date $DATE_TREE  --clock_rate ${SUB_TREE_CLOCK_RATE[$X]} --chain 300000000 # --cutoff 2.25 --grid_points $[ 225 * 52 / 100 ]
done

Executing BEAST MCMCs.

for X in $SUB_TREES; do
    #current jobs number are 31..35, previously it was 1..5 for all and 6..20 for B.1.1.7
    for i in {31..35}; do
        # For job sbumission: qsub -cwd -N beast-$X-$i -M foroughmand@gmail.com -l h_vmem=10G,mem_free=10G,s_vmem=10G -pe smp 3 -o results/beast/run/out/$X-$i -e results/beast/run/out/$X-$i run-beast.sh run/$X-$i $DATE_TREE $X
        # To be checked:
        bash run-beast.sh run/$X-$i $DATE_TREE $X
    done
done

Analysing the logs:

for X in $SUB_TREES; do
    for i in {31..35}; do
        loganalyser results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE.log
    done
done

Combining logs:

#combining and making a log file for each subtree
for X in $SUB_TREES; do
    logcombiner `
        for i in {31..35}; do
            echo results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE-fixed.log
        done` results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.log
done

Location assignment MCMC Step (DTA)

  • Input: results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE.trees results/gisaid-$DATE_TREE-metadata-sampled.tsv
  • Output: results/beast/run/$X-$i/ for $X in $SUB_TREES and $i in 31-32.
  • Intermediate files: results/beast/$X-DTA-$DATE_TREE.xml results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.trees

Resampling from sampled MCMC trees for next MCMC steps:

for X in $SUB_TREES; do
    MCMC_FRP_LOG_FILES=""
    RANGE=`seq 31 35`
    for i in $RANGE; do
        python scripts/resample.py --burnin 15000000 --rate 100000 --tree results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE.trees --out results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE-sampled.trees
        python scripts/resample.py --burnin 15000000 --count 500 --tree results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE.trees --out results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE-sub500.trees
        MCMC_FRP_LOG_FILES="$MCMC_FRP_LOG_FILES results/beast/run/$X-$i/$X.fixedRootPrior.skygrid-$DATE_TREE-sub500.trees"
    done
    logcombiner -trees $MCMC_FRP_LOG_FILES results/beast/$X.fixedRootPrior.skygrid-$DATE_TREE.trees
done

Generating XML files:

for X in $SUB_TREES; do
        ./scripts/fill-template.py --template data/X-DTA-template.xml --name $X --sample results/gisaid-$DATE_TREE-$X-samples1.txt --metadata results/gisaid-$DATE_TREE-metadata-sampled.tsv --out results/beast/$X-DTA-$DATE_TREE.xml --loc Germany --date $DATE_TREE
done

Executing BEAST MCMC:

for X in $SUB_TREES; do
    for i in {31..32}; do
        # Submission command: qsub -cwd -N DTA-$X-$i -l h_vmem=64G,mem_free=20G,s_vmem=20G -pe smp 5 -o results/beast/run/out/$X-$i -e results/beast/run/out/$X-$i run-beast-DTA.sh run/$X-$i $DATE_TREE $X
        # To be checked
        bash run/$X-$i $DATE_TREE $X
    done
done

Creating MCC tree:

  • Input: results/beast/run/$X-$i/$X-DTA-$DATE_TREE.sub4500.trees for $X in $SUB_TREES and $i in 31-32 results/beast/run/$X-$i/$X-DTA-$DATE_TREE.log
  • Output: results/beast/run/all/$X-DTA-$DATE_TREE.MCC.tree for $X in $SUB_TREES

Combining logs of DTA MCMC:

for X in $SUB_TREES; do
    MCMC_DTA_LOG_FILES=""
    MCMC_DTA_TREE_FILES=""
    for i in {31..32}; do
        python scripts/resample.py --burnin 500000 --rate 4500 --tree results/beast/run/$X-$i/$X-DTA-$DATE_TREE.trees --out results/beast/run/$X-$i/$X-DTA-$DATE_TREE.sub4500.trees
        MCMC_DTA_LOG_FILES="$MCMC_DTA_LOG_FILES results/beast/run/$X-$i/$X-DTA-$DATE_TREE.log"
        MCMC_DTA_TREE_FILES="$MCMC_DTA_TREE_FILES results/beast/run/$X-$i/$X-DTA-$DATE_TREE.sub4500.trees"
    done
    echo logcombiner -trees $MCMC_DTA_TREE_FILES results/beast/run/all/$X-DTA-$DATE_TREE.combined.trees
    logcombiner -trees $MCMC_DTA_TREE_FILES results/beast/run/all/$X-DTA-$DATE_TREE.combined.trees
    echo logcombiner $MCMC_DTA_LOG_FILES results/beast/run/all/$X-DTA-$DATE_TREE.combined.log
    logcombiner $MCMC_DTA_LOG_FILES results/beast/run/all/$X-DTA-$DATE_TREE.combined.log
done

Running tree-annotator from BEAST command-line tools.

mkdir results/beast/run/all
for i in {1..1}; do
    for X in $SUB_TREES; do
        # Job submission command: qsub -cwd -N ta-$X-$i -l h_vmem=64G,mem_free=20G,s_vmem=20G -pe smp 2 -o results/beast/run/out/$X-$i -e results/beast/run/out/$X-$i run-treeannotator.sh $DATE_TREE $X $i
        # To be checked
        bash run-treeannotator.sh $DATE_TREE $X $i
    done
done

(Optional) Compressing log files and sampled trees.

for X in $SUB_TREES; do  
    xz -f -k -v -T0 results/beast/run/all/$X-DTA-$DATE_TREE.combined.trees
    xz -f -v -T0 results/beast/run/all/$X-DTA-$DATE_TREE.combined.log
done

Generating reports for MCC tree for sub-sampled sequences (Optional)

For the sub-sampled sequences not the results are at ../results/beast/run/all/ (relative to the working directory). The reports on the reports folder could be executed and lineages for this tree could be generated and evaluated.

Extracting importation lineages and adding unsampled sequences from Germany to the importation lienages

  • Input: results/gisaid-$DATE_TREE-samples.txt results/gisaid-$DATE_TREE-unsampled.txt results/gisaid-$DATE_TREE-unsampled-subtree.txt ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv ../../../data/data/mmsa_20210622_masked.fa
  • Output: results/beast/run/lin-ius/clusters_DTA_MCC_NA.tsv results/beast/run/lin-ius/clusterSamples_DTA_MCC_NA.tsv results/beast/run/lin-ius/clusters_DTA_MCC_0.5.tsv results/beast/run/lin-ius/clusterSamples_DTA_MCC_0.5.tsv
  • Intermediate files: results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_NA.tsv results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_NA.tsv results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_singles.tsv results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_0.5.tsv results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_0.5.tsv results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_singles_0.5.tsv
scripts/extract-unsampled --location_label $STATE non$STATE --in results/gisaid-$DATE_TREE-$STATE.tree --samples results/gisaid-$DATE_TREE-samples.txt --unsampled results/gisaid-$DATE_TREE-unsampled.txt $SUB_TREES --cond 2 "==" Germany --metadata ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv
cat results/gisaid-$DATE_TREE-unsampled.txt | grep -v "NA$" > results/gisaid-$DATE_TREE-unsampled-subtree.txt
./scripts/extract-metadata.R results/gisaid-$DATE_TREE-unsampled-subtree.txt ../../../data/data/gisaid-$DATE_METADATA-metadata.tsv results/gisaid-$DATE_TREE-metadata-unsampled.tsv
cat results/gisaid-$DATE_TREE-metadata-sampled.tsv results/gisaid-$DATE_TREE-metadata-unsampled.tsv  > results/gisaid-$DATE_TREE-metadata-sampled-unsampled.tsv

Creating alignment file for unsampled sequences from Germany.

cut -f3 results/gisaid-$DATE_TREE-metadata-unsampled.tsv | tail -n +2  > results/mmsa-$DATE_TREE-unsample-ids.txt
for X in $SUB_TREES; do cat results/gisaid-$DATE_TREE-$X-samples0.txt; done > results/gisaid-$DATE_TREE-all-samples0.txt
./scripts/alignment-filter ../../../data/data/mmsa_20210622_masked.fa <(cat results/mmsa-$DATE_TREE-unsample-ids.txt results/gisaid-$DATE_TREE-all-samples0.txt) > ../../data/phylogenetic/mmsa-$DATE_SEQ-sampled-unsampled0.fa

for X in $SUB_TREES; do
        echo "Enriching tree $X"
        cp results/beast/run/all/$X-DTA-$DATE_TREE.MCC.tree results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.nexus
        ./scripts/phylogeo_sankoff_general --in results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.nexus --out results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree --metadata results/gisaid-$DATE_TREE-metadata-unsampled.tsv --location_label \"$STATE\" \"non$STATE\" --cond 2 "==" Germany --nosankoff --print-allow-annotation false --single-child false --ilabel true --set-tip-location false
        ./scripts/phylogeo_sankoff_general --in results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.nexus --out results/beast/unsampled/$X-DTA-$DATE_TREE-rich.MCC.tree --metadata results/gisaid-$DATE_TREE-metadata-unsampled.tsv --location_label \"$STATE\" \"non$STATE\" --cond 2 "==" Germany --nosankoff --print-allow-annotation true --single-child false --ilabel true --set-tip-location false

        ./scripts/alignment-filter ../../data/phylogenetic/mmsa-$DATE_SEQ-sampled-unsampled0.fa <( grep "$X$" results/gisaid-$DATE_TREE-unsampled-subtree.txt | cut -f 1 ) > results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query-mmsa.fa
        ./scripts/alignment-filter ../../data/phylogenetic/mmsa-$DATE_SEQ-sampled-unsampled0.fa results/gisaid-$DATE_TREE-$X-samples0.txt > results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference-mmsa.fa
done

Some initializations:

for X in $SUB_TREES; do
        mv results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference-mmsa.fasta
        mv results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query-mmsa.fasta
        mv results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference-mmsa.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference-mmsa.fa
        mv results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query-mmsa.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query-mmsa.fa
done
mkdir results/beast/run/lin-ius/ results/beast/run/lin-ius/out/
mkdir results/beast/run/lin-ius/aln/ results/beast/run/lin-ius/alnq/
for X in $SUB_TREES; do
        echo "Tree $X "
        #calculate distances between sequences
        csplit -q --prefix=results/beast/run/lin-ius/aln/alignment-$DATE_TREE-$X-epa-reference- --suffix-format=%06d.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/reference-mmsa.fa '/^>/' '{*}'
        rm results/beast/run/lin-ius/aln/alignment-$DATE_TREE-$X-epa-reference-000000.fasta
        csplit -q --prefix=results/beast/run/lin-ius/alnq/alignment-$DATE_TREE-$X-epa-query- --suffix-format=%06d.fasta results/beast/unsampled/alignment-$DATE_TREE-$X-epa/query-mmsa.fa '/^>/' '{*}'
        rm results/beast/run/lin-ius/alnq/alignment-$DATE_TREE-$X-epa-query-000000.fasta
        ls -S results/beast/run/lin-ius/aln/ | grep "alignment-$DATE_TREE-$X-epa-reference-" | awk '{print "'results/beast/run/lin-ius/aln/'" $0}' | grep -v '\-000000.fasta' > results/beast/run/lin-ius/out/alignment-files-$X-reference.txt
        ls -S results/beast/run/lin-ius/alnq/ | grep "alignment-$DATE_TREE-$X-epa-query-" | awk '{print "'results/beast/run/lin-ius/alnq/'" $0}' | grep -v '\-000000.fasta' > results/beast/run/lin-ius/out/alignment-files-$X-query.txt
        mash sketch -o results/beast/run/lin-ius/alignment-$DATE_TREE-$X-epa-reference.fasta results/beast/run/lin-ius/aln/alignment-$DATE_TREE-$X-epa-reference-*.fasta  2>>e
        mash sketch -o results/beast/run/lin-ius/alignment-$DATE_TREE-$X-epa-query.fasta -l results/beast/run/lin-ius/out/alignment-files-$X-query.txt 2>>e
        time mash dist -v 0.05 -p 120 results/beast/run/lin-ius/alignment-$DATE_TREE-$X-epa-reference.fasta.msh results/beast/run/lin-ius/alignment-$DATE_TREE-$X-epa-query.fasta.msh | ~/bin/pigz-2.6/pigz -k -3 -p10 > results/beast/run/lin-ius/alignment-distance-$DATE_TREE-$X.txt.gz
        zcat results/beast/run/lin-ius/alignment-distance-$DATE_TREE-$X.txt.gz | scripts/convert-filename-to-id | gzip > results/beast/run/lin-ius/alignment-distance-$DATE_TREE-$X-id.txt.gz

        sed 's/Germany+nonGermany/nonGermany/g' results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.nexus > results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.2.nexus
        scripts/lineage-importation-inject-unsampled --tree results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.2.nexus --dist <( zcat results/beast/run/lin-ius/alignment-distance-$DATE_TREE-$X-id.txt.gz ) -l \"$STATE\" \"non$STATE\" --out-folder results/beast/run/lin-ius/ --metadata results/gisaid-$DATE_TREE-metadata-sampled-unsampled.tsv --lin-prefix LIN-Germany-$X-"$DATE_TREE"_DTA_MCC_ --treefile results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.nexus --out-clusters results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_NA.tsv --out-cluster-samples results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_NA.tsv --out-clusters-single results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_singles.tsv --out-0.5 results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_0.5.tsv results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_0.5.tsv results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_singles_0.5.tsv

done

Creating output files (to be used by R scripts for generating the reports):

(
X=`echo $SUB_TREES | head -n1 | awk '{print $1;}'`
head -n 1 results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_NA.tsv
for X in $SUB_TREES; do
        tail -n +2 results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_NA.tsv 
done ) > results/beast/run/lin-ius/clusters_DTA_MCC_NA.tsv

(
X=`echo $SUB_TREES | head -n1 | awk '{print $1;}'`
head -n 1 results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_NA.tsv
for X in $SUB_TREES; do
        tail -n +2 results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_NA.tsv
done ) > results/beast/run/lin-ius/clusterSamples_DTA_MCC_NA.tsv

(
X=`echo $SUB_TREES | head -n1 | awk '{print $1;}'`
head -n 1 results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_0.5.tsv
for X in $SUB_TREES; do
        tail -n +2 results/beast/run/lin-ius/out/$X-clusters_DTA_MCC_0.5.tsv 
done ) > results/beast/run/lin-ius/clusters_DTA_MCC_0.5.tsv

(
X=`echo $SUB_TREES | head -n1 | awk '{print $1;}'`
head -n 1 results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_0.5.tsv
for X in $SUB_TREES; do
        tail -n +2 results/beast/run/lin-ius/out/$X-clusterSamples_DTA_MCC_0.5.tsv
done ) > results/beast/run/lin-ius/clusterSamples_DTA_MCC_0.5.tsv

Internal movement analysis

The internal movement analysis is a Bayesian DTA method that assigns location, state of Germany, to internal nodes of each imported lineage. This is done on the main folders for each subsampling method, e.g. analyses/phylogenetic-test-snake-2/, analyses/phylogenetic-test-subsampling-3/, and analyses/phylogenetic-test-subsampling-5/.

First, we remove nodes with unknown state.

for X in $SUB_TREES; do
  ./scripts/metadata-remove-na.R results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv results/dtamulti/gisaid-$DATE_TREE-$X-metadata-sampled-no-na.tsv
  ./scripts/phylogeo_sankoff_general --in  results/beast/unsampled/$X-DTA-$DATE_TREE.MCC.tree.2.nexus --save-nexus --out results/dtamulti/$X-DTA-$DATE_TREE.MCC.tree.2.nexus --metadata results/gisaid-$DATE_TREE-$X-metadata-sampled.tsv --location_label \"$STATE\" \"non$STATE\" --cond 2 "==" Germany  --print-annotation false --print-internal-node-label false --single-child false --nosankoff --remove-invalid-children true
done

for X in $SUB_TREES; do
  cat results/dtamulti/gisaid-$DATE_TREE-$X-metadata-sampled-no-na.tsv
done > results/dtamulti/gisaid-$DATE_TREE-all-metadata-sampled-no-na.tsv

Then, we merge all the imported lineages to one tree with a root with long branches. This helps us in having one set of parameters for all the different lienages.

./scripts/merge-trees-for-multidta --trees `for X in $SUB_TREES; do echo results/dtamulti/$X-DTA-$DATE_TREE.MCC.tree.2.nexus; done` --root-date 1900 -l \"Germany\" \"nonGermany\" --metadata results/gisaid-$DATE_TREE-metadata-sampled.tsv --metadata-in-filter results/dtamulti/gisaid-$DATE_TREE-all-metadata-sampled-no-na.tsv --samples results/dtamulti/sampled.fixedRootPrior.skygrid-$DATE_TREE.samples --out results/dtamulti/sampled.fixedRootPrior.skygrid-$DATE_TREE.trees

Then, we fill the template and execute the MCMC via BEAST.

./scripts/fill-template.py --template data/X-DTAmulti-template.xml --name sampled --sample results/dtamulti/sampled.fixedRootPrior.skygrid-$DATE_TREE.samples --metadata results/gisaid-$DATE_TREE-metadata-sampled.tsv --out results/dtamulti/sampled-DTAmulti-$DATE_TREE.0.xml --loc Germany --date $DATE_TREE --loc_depth

sed 's/Baden-Württemberg/Baden-Wurttemberg/g' < results/dtamulti/sampled-DTAmulti-$DATE_TREE.0.xml > results/dtamulti/sampled-DTAmulti-$DATE_TREE.xml

for i in {31..32}; do
  bash run-beast-DTAmulti.sh run/sampled-$i $DATE_TREE sampled
done

Then, we merge the trees from two independent runs and subsample them for tree annotator.

mkdir results/dtamulti/run/all results/dtamulti/run/lin-ius-3/
MCMC_DTA_LOG_FILES=""
MCMC_DTA_TREE_FILES=""
for i in {31..32}; do
srun python scripts/resample.py --burnin 500000 --rate 4500 --tree results/dtamulti/run/sampled-$i/sampled-DTA-$DATE_TREE.trees --out results/dtamulti/run/sampled-$i/sampled-DTA-$DATE_TREE.sub4500.trees
MCMC_DTA_LOG_FILES="$MCMC_DTA_LOG_FILES results/dtamulti/run/sampled-$i/sampled-DTA-$DATE_TREE.log"
MCMC_DTA_TREE_FILES="$MCMC_DTA_TREE_FILES results/dtamulti/run/sampled-$i/sampled-DTA-$DATE_TREE.sub4500.trees"
done
srun logcombiner -trees $MCMC_DTA_TREE_FILES results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.trees
srun logcombiner $MCMC_DTA_LOG_FILES results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.log

loganalyser results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.log > results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.log.analyzed

Running tree annotator that finds MCC tree.

srun --mem=100G --cpus-per-task=2 --qos verylong --time=24:00:00 ../../bin/beast/bin/treeannotator -type mcc results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.trees  results/dtamulti/run/all/sampled-DTA-$DATE_TREE.MCC.nexus

Finally, the lineages will be extracted for the R report generation scripts

srun ./scripts/lineage-importation-extract-multidta --tree results/dtamulti/run/all/sampled-DTA-$DATE_TREE.MCC.nexus --lineage-samples results/beast/run/lin-ius-3/clusterSamples_DTA_MCC_0.5.tsv --out results/dtamulti/run/lin-ius-3/clusterMovement_DTA_MCC_0.5.0.tsv -l \"Baden-Wurttemberg\" \"Bavaria\" \"Berlin\" \"Brandenburg\" \"Bremen\" \"Hamburg\" \"Hesse\" \"Lower Saxony\" \"Mecklenburg-Western Pomerania\" \"North Rhine-Westphalia\" \"Rhineland-Palatinate\" \"Saarland\" \"Saxony\" \"Saxony-Anhalt\" \"Schleswig-Holstein\" \"Thuringia\" --root-date 1900
sed 's/"//g' < results/dtamulti/run/lin-ius-3/clusterMovement_DTA_MCC_0.5.0.tsv > results/dtamulti/run/lin-ius-3/clusterMovement_DTA_MCC_0.5.tsv

Generating the reports

  • Input: results/beast/run/lin-ius/clusters_DTA_MCC_NA.tsv results/beast/run/lin-ius/clusterSamples_DTA_MCC_NA.tsv results/beast/run/lin-ius/clusters_DTA_MCC_0.5.tsv results/beast/run/lin-ius/clusterSamples_DTA_MCC_0.5.tsv
  • Output: reports-ius/importationSummaryMultiState.pdf reports-ius/lineageBreakdownMultiState.pdf reports-ius/lineageSummaryMultiState.pdf

The R reports (Rmd files) on the folder reports-ius/ (relative to working directory) could be executed then. The reports use the files generated from the last step (importation lineage extraction and adding unsampled sequences).

Robustness analysis

We checked robustness of results under different sub-sampling methods.

Bayesian uncertainty analysis

In the main part of the study we inferred the importation lineages based on the MCC tree afthe the DTA Bayesian step. This could also be done directly based on the posterior samples. Here, for each posterior sample (tree and the location annotation), the importation lineages are inferred. The same methodology also used for the inter-state movement analysis of the importation lineages. The results are available as follows:

Uncertainty analysis of the importation lineages

Following codes infers the importation lineage for each Bayesian posterior sampled tree.

rm -rf results/beast/run/lin-ius-3e/

#extract trees results/beast/run/all/$X-DTA-$DATE_TREE.combined.trees to results/beast/run/lin-ius-3e/sample-tree
mkdir -p results/beast/run/lin-ius-3e/out results/beast/run/lin-ius-3e/sample-tree results/beast/run/lin-ius-3e/log results/beast/run/lin-ius-3e/mcc-tree/  results/beast/run/lin-ius-3e/out-tree/ 
for X in $SUB_TREES; do
	echo $X ...
	python scripts/tree-separate.py results/beast/run/all/$X-DTA-$DATE_TREE.combined.trees results/beast/run/lin-ius-3e/sample-tree/$X-NAME.nexus
done

for TREE_FILE in results/beast/run/lin-ius-3e/sample-tree/*.nexus ; do
TREE_NAME=`echo $(basename $TREE_FILE) | sed 's/\.nexus//'`
echo $TREE_NAME $TREE_FILE
bash scripts/generate-effectiveness.sh $DATE_TREE $STATE $TREE_NAME
done
# results in results/beast/run/lin-ius-3e/out/$TREE_NAME-effectiveness.tsv

Inferred importation lineages are calculated for each tree in the following code.

TREE_NAME=A-1000440000
head -n 1 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_0.5.tsv | sed 's/$/\ttree/' > results/beast/run/lin-ius-3e/out/all-clusters_DTA_MCC_0.5.tsv 
head -n 1 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_0.5.tsv | sed 's/$/\ttree/' > results/beast/run/lin-ius-3e/all-clusters_DTA_MCC_0.5.tsv
head -n 1 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusterSamples_DTA_MCC_0.5.tsv| sed 's/$/\ttree/' > results/beast/run/lin-ius-3e/out/all-clusterSamples_DTA_MCC_0.5.tsv
head -n 1 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_singles_0.5.tsv | sed 's/$/\ttree/' >> results/beast/run/lin-ius-3e/out/all-clusters_DTA_MCC_singles_0.5.tsv

for TREE_FILE in results/beast/run/lin-ius-3e/sample-tree/*.nexus ; do
TREE_NAME=`echo $(basename $TREE_FILE) | sed 's/\.nexus//'`
#echo $TREE_NAME $TREE_FILE
tail -n+2 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_0.5.tsv | sed 's/$/\t'$TREE_NAME'/' >> results/beast/run/lin-ius-3e/out/all-clusters_DTA_MCC_0.5.tsv 
tail -n+2 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_0.5.tsv | sed 's/$/\t'$TREE_NAME'/' >> results/beast/run/lin-ius-3e/all-clusters_DTA_MCC_0.5.tsv
tail -n+2 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusterSamples_DTA_MCC_0.5.tsv | sed 's/$/\t'$TREE_NAME'/' >> results/beast/run/lin-ius-3e/out/all-clusterSamples_DTA_MCC_0.5.tsv
tail -n+2 results/beast/run/lin-ius-3e/out/$TREE_NAME-clusters_DTA_MCC_singles_0.5.tsv | sed 's/$/\t'$TREE_NAME'/' >> results/beast/run/lin-ius-3e/out/all-clusters_DTA_MCC_singles_0.5.tsv
done

The following Rscript summarizes the calculated effectiveness values for the trees and calculates the 95% HPDs.

library(dplyr)
library(tidyr)
library(bfp)
d <- read.table('results/beast/run/lin-ius-3e/all.tsv', sep='\t', header=TRUE, stringsAsFactor=FALSE)
colnames(d) <- c('name', 'N1', 'N2', 'N3', 'N4', 'N5', 'N6', 'N7', 'N8',  'N9', 'N10', 'N11', 'N12', 'tree')
d<- d %>% filter(name != 'name')
ds <- d %>% mutate(name = ifelse(name == " no NRW,BV", "no NRW,BV", name)) %>% pivot_longer(num_range("N", 1:12), names_to='npi', values_to='eff') %>% mutate(eff = as.numeric(eff)) %>% group_by(name, npi) %>% summarise(eff.avg = mean(eff), eff.sd=sd(eff), hpd.l = empiricalHpd(eff, level=0.95)[1], hpd.u=empiricalHpd(eff, level=0.95)[2]) %>% pivot_wider(names_from=npi, values_from=c(eff.avg, eff.sd, hpd.l, hpd.u)) %>% select(sort(tidyselect::peek_vars())) %>% relocate(name) 
write.table(ds, 'results/beast/run/lin-ius-3e/summ.tsv', sep='\t', quote=FALSE, row.names=FALSE)

Uncertainty analysis of the inter-state movement of the importation lineages

The inter-state movement of the importation lineages could be inferred based on the posterior samples produced with the Bayesian method.

rm -rf results/dtamulti/run/lin-ius-3e
mkdir -p results/dtamulti/run/lin-ius-3e/sample-tree/ results/dtamulti/run/lin-ius-3e/movement/ results/dtamulti/run/lin-ius-3e/log/

python scripts/tree-separate.py results/dtamulti/run/all/sampled-DTA-$DATE_TREE.combined.trees results/dtamulti/run/lin-ius-3e/sample-tree/tree-NAME.single.nexus

for TREE_FILE in results/dtamulti/run/lin-ius-3e/sample-tree/*.single.nexus ; do
TREE_NAME=`echo $(basename $TREE_FILE) | sed 's/\.single\.nexus//'`
echo $TREE_NAME $TREE_FILE
bash run-2-movement.sh $TREE_NAME
done

TREE_NAME=tree-997416000
head -n 1 results/dtamulti/run/lin-ius-3e/movement/$TREE_NAME.clusterMovement_DTA_MCC_0.5.0.tsv | sed 's/$/\ttree/' > results/dtamulti/run/lin-ius-3e/all.clusterMovement_DTA_MCC_0.5.0.tsv

for TREE_FILE in results/dtamulti/run/lin-ius-3e/sample-tree/*.single.nexus ; do
TREE_NAME=`echo $(basename $TREE_FILE) | sed 's/\.single\.nexus//'`
tail -n+2 results/dtamulti/run/lin-ius-3e/movement/$TREE_NAME.clusterMovement_DTA_MCC_0.5.0.tsv | sed 's/$/\t'$TREE_NAME'/' >> results/dtamulti/run/lin-ius-3e/all.clusterMovement_DTA_MCC_0.5.0.tsv
done

About

SARS-CoV-2 lineage importations and spread are reduced after nonpharmaceutical interventions in phylogeographic analyses

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages