HGENII Module3 Project
- HGEN report revised.pdf
This the write up which includes results and discussion of the results.
- proj_Ling.py
This is the main script for calculation allele frequencies, hwe, ld, PCA and inferring subpopulation membership. The usage of this script is listed below. Basically, it takes a given data table (in the exact format as suggested in the project description) with -i option and compute the answers to all questions of the project3 with --all option or selectively compute the answer of some of the questions as suggested below.
./proj_Ling.py -h
usage: proj_Ling.py [-h] -i INPUT [-f] [--hwe] [--ld] [--pca]
[--subpopmembership] [-k K] [--nofvariants NOFVARIANTS]
[--rounds ROUNDS] [--all] -o OUT
HGEN module 3 homework
optional arguments:
-h, --help show this help message and exit
-i INPUT, --input INPUT
input the data table which should be an N by M+2
table, with each row representing an individual, and
the 1st column is ID, 2nd affection status, remaining
columns genotypes
-f, --frequency Plot a histogram of the allele frequency
--hwe Calculate the p value of HWE for each variant and plot
the distribution of the p values and QQ plot of the p
values (log scale)
--ld Calculate and plot 3 heatmaps of pairwise LD among all
pairs of the variants: D, D', r2
--pca Perform PCA and plot PC1 vs PC2, PC1 vs PC3, PC2 vs
PC3
--subpopmembership Calculate the membership of each individual with the
marginal probability of being in the assined
substructure based on the Gibbs sampler results,
default K=2, use the common variants in the first 1000
variants, 200 rounds
-k K Specify the number of populations to infer from the
gibbs sampling method
--nofvariants NOFVARIANTS
Specify the number of variants to perform the gibbs
sampling method
--rounds ROUNDS Specify the number of rounds for the gibbs sampling
method to reach a stationary distribution
--all Do all the analysis, generate a pdf report
-o OUT, --out OUT The file name of the pdf report
One caveat is that when calculating the LD, it only computes for polymorphic variants of the first 500 variants of each dataset due to the computation power.
Required packages: numpy, pandas, matplotlib, seaborn, sklearn, scipy
Example 1: If I want to compute the subpopulation membership of the data, I will input:
./proj_Ling.py -i data.table.txt --subpopmembership -k 2 --nofvariants 1000 --rounds 200 -o test.pdf
Example 2: If I want to compute all the questions of the data, I will input:
./proj_Ling.py -i data.table.txt --all -o test.pdf
- HGEN341_module3_final_revised.ipynb/html
This jupyter notebook have the code, answers and the plots. The html is a static version of the notebook.