gnames is a Python 3.x package for efficient simulation of multigenerational data under assortative mating and genetic nurture effects.
gnames, please make sure Git and Anaconda with Python 3.x are installed.
In order to download gnames, open a command-line interface by starting Anaconda Prompt, navigate to your working directory, and clone the gnames repository using the following command:
git clone https://github.com/devlaming/gnames.git
Now, enter the newly created gnames directory using:
cd gnames
Then run the following commands to create a custom Python environment which has all of gnames's dependencies (i.e. an environment that has packages numpy, pandas, scipy, and tqdm pre-installed):
conda env create --file gnames.yml
conda activate gnames
(or activate gnames instead of conda activate gnames on some machines).
In case you cannot create a customised conda environment (e.g. because of insufficient user rights) or simply prefer to use Anaconda Navigator or pip to install packages e.g. in your base environment rather than a custom environment, please note that gnames only requires Python 3.x with the packages numpy, pandas, scipy, and tqdm installed.
You can now run the following commands, to test if gnames is functioning properly:
python -c "from gnames import gnames; gnames.Test()"
This command should yield output along the following lines:
TEST OF GNAMES
with 1000 founders, 10,000 SNPs, and two children per pair
INITIALISING SIMULATOR
Drawing alleles for SNPs of founders
Drawing allele frequencies for SNPs of founders
Drawing true SNP effects
Drawing genotypes founders
Highest diagonal element of GRM for founders = 1.048
SIMULATING 10 GENERATIONS
100%|█████████████████████████| 10/10 [00:03<00:00, 2.72it/s]
Highest diagonal element of GRM after 10 generations = 1.065
GENERATING OUTPUT
Calculating and storing classical GWAS and within-family GWAS
results based on offspring data last generation
Writing PLINK files (genotypes.bed,.bim,.fam,.phe)
Making GRM in GCTA binary format
(genotypes.grm.bin,.grm.N.bin,.grm.id)
Making 3 PGIs in hold-out sample based on 3 sets of
GWAS estimates (GWAS 1 & 2: non-overlapping; GWAS 3: pooled;
all sampling 1 child per family)
Runtime: 4.271 seconds
This output shows gnames simulated a founder population comprising 1000 individuals and 10,000 SNPs. Subsequently, gnames simulated ten generations of offspring data under genetic nurture and assortative mating. gnames reports that the highest element of the diagonal of the GRM increased from 1.048 to 1.066 over the ten generations.
In addition, gnames performed a classical GWAS and a within-family GWAS based on the offspring data for the last generation. Results are exported to human-readable files: results.GWAS.classical.txt and results.GWAS.within_family.txt.
Moreover, gnames created a set of PLINK binary files: genotypes.bed, genotypes.bim, genotypes.fam. These PLINK binary files can readily be used for follow-up analyses using tools such as PLINK. gnames also created a phenotype file, genotypes.phe, that can be used by PLINK e.g. to perform a GWAS.
Also, gnames created a set of GRM files in GCTA binary format: genotypes.grm.id, genotypes.grm.bin, and genotypes.grm.N.bin. These files combined with genotypes.phe can easily be used for follow-up analyses using tools such as MGREML and GCTA.
Finally, gnames performed three classical GWASs (i) using data on only child per family (ii) for two non-overlapping sets of families in GWAS 1 and 2 and (iii) for these two GWAS samples pooled in GWAS 3, where (iv) 40% of the families are considered in GWAS 1 and 2, and, thus, 80% in GWAS 3. These 3 sets of GWAS results are used for polygenic prediction out-of-sample for all children for the remaining 20% of the families. Resulting polygenic scores or polygenic indices (PGIs) can be found together with the true phenotype Y, its genetic component G, its environment component E, and its genetic nurture component N in results.pgi.
The additional output file results.info shows the number of SNPs contributing directly to Y (as SNPs with minor allele frequency equal to zero only count towards the intercept of Y and, hence, do not contribute to variation in Y directly) and the number of SNPs used to construct PGIs (SNPs with MAF equal to zero in either of the GWAS samples are excluded). You might wonder: why can the effective number of markers differ from the number of markers that gnames started with? The reason here is quite simple: gnames simulates data on multiple generations. Therefore, MAFs evolve over generations and can, thus, enter the absorbing state where the MAF is zero.
The whole simulation, four GWASs, polygenic prediction, and export to PLINK binary files and to GCTA binary GRM files took less than five seconds.
Once gnames is up-and-running, you can simply incorporate the tool in your Python code, as illustrated in the following bit of Python code:
from gnames import gnames
import numpy as np
import matplotlib.pyplot as plt
N=1000
M=10000
T=1000
F=0.4
gsimulator=gnames(N,M,dMAF0=F)
vDiags0=np.sort(gsimulator.ComputeDiagsGRM())
gsimulator.Simulate(T)
vDiags1000=np.sort(gsimulator.ComputeDiagsGRM())
plt.plot(np.vstack((vDiags0,vDiags1000)).T)
plt.savefig('diagsGRM.pdf')
plt.close()
gsimulator.PerformGWAS('n1000.m10000.t1000')
gsimulator.MakeBed('n1000.m10000.t1000')
gsimulator.MakeGRM('n1000.m10000.t1000')
The plot that is created near the end of the code shows the diagonal elements of the GRM sorted from small to large for the founders (blue line) and for the 1000th offspring generation (orange line). As a result of considerable assortative mating in this simulation, we can see that the diagonal elements of the GRM have considerably shifted away from one over the generations.
In addition, this bit of code shows how gnames can be used to calculate GWAS summary statistics based on the last generation, here yielding files named n1000.m10000.t1000.GWAS.classical.txt and n1000.m10000.t1000.GWAS.within_family.txt.
Finally, the code also shows how gnames can be used to create PLINK binary files and binary GRM files for the last generation. These files are here named (i) n1000.m10000.t1000.bed, n1000.m10000.t1000.bim, and n1000.m10000.t1000.fam and (ii) n1000.m10000.t1000.grm.id, n1000.m10000.t1000.grm.bin, and n1000.m10000.t1000.grm.N.bin.
You can update to the newest version of gnames using git. First, navigate to your gnames directory (e.g. cd gnames), then run
git pull
If gnames is up to date, you will see
Already up to date.
otherwise, you will see git output similar to
remote: Enumerating objects: 8, done.
remote: Counting objects: 100% (8/8), done.
remote: Compressing objects: 100% (4/4), done.
remote: Total 6 (delta 2), reused 6 (delta 2), pack-reused 0
Unpacking objects: 100% (6/6), 2.82 KiB | 240.00 KiB/s, done.
From https://github.com/devlaming/gnames
481a4bf..fddd8cc main -> origin/main
Updating 481a4bf..fddd8cc
Fast-forward
README.md | 128 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
gnames.py | 26 ++++++++++++-
2 files changed, 153 insertions(+), 1 deletion(-)
create mode 100644 README.md
which tells you which files were changed.
If you have modified the gnames source code yourself, git pull may fail with an error such as error: Your local changes [...] would be overwritten by merge.
Before contacting me, please try the following:
- Go over the tutorial in this
README.mdfile - Go over the method, described in tba (citation below)
In case you have a question that is not resolved by going over the preceding two steps, or in case you have encountered a bug, please send an e-mail to r[dot]devlaming[at]vu[dot]nl.
If you use the software, please cite the manuscript in which gnames was first described and utilised:
This project is licensed under GNU GPL v3.
Ronald de Vlaming (Vrije Universiteit Amsterdam)