# Simulate 3-way admixture issues

[rpauly]

Hi, I am trying the command: admix haptools-simu-admix \ --pfile data/example_data/1kg-ref \ --admix-prop '{"CEU": 0.4, "YRI": 0.1, "PEL": 0.5}' \ --pop-col Population \ --mapdir data/1kg-ref-${BUILD}/metadata/genetic_map/ \ --n-gen 10 \ --n-indiv 10000 \ --out data/example_data/CEU-YRI-PEL

But, I get an error: ERROR: Cannot find key: haptools-simu-admix Usage: admix <group|command> available groups: fire available commands: assoc | append_snp_info | calc_pgs | calc_partial_pgs | grm | log_params | get_1kg_ref | select_admix_indiv | simulate_admix_pheno | simulate_pheno | lanc | lanc_convert | lanc_count | prune | pca | liftover | pfile_align_snp | pfile_merge_indiv | pfile_freq_within | subset_hapmap3 | subset_pop_indiv | hapgen2 | admix_simu | download_dependency | plot_joint_pca | admix_grm | admix_grm_merge | genet_cor | admix_grm_rho | estimate_genetic_cor | summarize_genet_cor | meta_analyze_genet_cor | cli

For detailed information on this command, run: admix --help

[KangchengHou]

Thank you for your interest!

Recently we having been updating the software and forgot to update the pip source.

Could you try pip install -U admix-kit and try the command again? Thanks

[rpauly]

Can I use the conda package instead, using below: conda env create --file environment.yaml conda activate admix-kit

[KangchengHou]

Yes this is another option - just making sure that you are cloning the master branch

[rpauly]

Thanks another question, does this mean that each simulated individual will have CEU: 40%, YRI: 10% and PEL: 50%?

[KangchengHou]

No, these parameters mean that the simulated individuals will have these proportions on average, but the ancestry proportion for each individual may vary.

[rpauly]

thank you...so if admix-prop '{"CEU": 0.2, "YRI": 0.8}'...all the samples would have a higher proportion of YRI proportions correct? I am asking because, I am trying to simulate some data and I am trying to get individuals with the same admix proportions.

[KangchengHou]

Yes the samples would have a overall higher proportion of YRI proportions - although under rare cases you may encounter some individuals with YRI < 50% purely due to the randomness.

You can simulate a bunch of individuals and then filter based on ancestry proportions afterwards

[rpauly]

thank you...that is very interesting...is there a command I can use to filter the individuals based on ancestry proportions?

[KangchengHou]

hey unfortunately there is not a command to filter - you have to manually calculate genetic ancestry and then filter by indexing those selected individuals

[rpauly]

Thanks...so I was trying to run all the chromosomes through the pipeline and I got the first two steps: Step1:

for CHROM in {9..22}; do
source admix_env/bin/activate
admix subset-hapmap3 --pfile /data/1kg-ref-hg38/pgen/all_hg38 --build hg38 --chrom ${CHROM} --out hm3_chrom${CHROM}.snp
done

Step2:

for CHROM in {10..22}; do
plink2 --pfile /data/1kg-ref-hg38/pgen/all_hg38 --extract hm3_chrom${CHROM}.snp --make-pgen --out 1kg-ref/1kg-ref_chrom${CHROM} --allow-extra-chr
done

Step3:

admix haptools-simu-admix \
    --pfile ${OUT_DIR}/1kg-ref \
    --admix-prop '{"CEU": 0.2, "YRI": 0.8}' \
    --pop-col Population \
    --mapdir data/1kg-ref-${BUILD}/metadata/genetic_map/ \
    --n-gen 10 \
    --n-indiv 10000 \
    --out ${OUT_DIR}/CEU-YRI

But Step3 gives me an error in the mapdir step, How do I generate genetic maps? Thanks, Rini

[KangchengHou]

Could you type ls data/1kg-ref-${BUILD}/metadata to see what is in there? It should be already downloaded using the get-1kg-ref command.

[rpauly]

The metadata is blank

[KangchengHou]

Ok can you try this to download the genetic map

root_dir=data/1kg-ref-hg38
mkdir -p ${root_dir}/metadata/genetic_map/raw && cd ${root_dir}/metadata/genetic_map/raw
wget https://bochet.gcc.biostat.washington.edu/beagle/genetic_maps/plink.GRCh38.map.zip
unzip plink.GRCh38.map.zip

cd ../../../../../

# move raw files to out/metadata/genetic_map
for chrom in {1..22}; do
    cp ${root_dir}/metadata/genetic_map/raw/plink.chr${chrom}.GRCh38.map ${root_dir}/metadata/genetic_map/chr${chrom}.map
done

# clean up
rm -rf ${root_dir}/metadata/genetic_map/raw/

[rpauly]

Awesome that works, so for: step3: admix haptools-simu-admix --pfile 1kg-ref/1kg-ref_chrom10 --admix-prop '{"CEU": 0.2,"YRI": 0.8}' --pop-col Population --mapdir data/1kg-ref/metadata/genetic_map/ --n-gen 10 --n-indiv 11 --out data/example_data/CEU-YRI_n_gene_simulations_n_gen_5

In the above command, I have used only chromsome 10. Is there a way I can combine all the chromosomes pgen/pvar/psam into a single pgen/pvar/psam file and run the above command just once? Thanks!

[KangchengHou]

Unfortunately you have to run jobs seperately for each chromosome. Running in parallel will speed up your analysis if you have access to a computing cluster

[rpauly]

Thanks for all your help! It works. One last question, is there a way to calculate the genetic ancestry on the admixed sample? For instance, I saw that " .af_per_anc()" calculates allele frequencies per ancestry backgrounds. How can I use this ?

[KangchengHou]

Apology for the late reply.

.af_per_anc() is used for calculating the local ancestry-specific alllele frequency.

You can use admix lanc-count --lanc [prefix].lanc --out [prefix].lanc_count to compute the number of alleles from each genetic ancestry