These are scripts for running a local eQTL and aseQTL analysis.
The main idea is that you want to keep these scripts in a bin folder and create a "data" folder and a "results" folder adjacent to your scripts folder. Inside the results folder, mmake specific folders for each run -- I do this using dates.
To use these you will need to install SciPy and NumPy
Here are the steps for analysis:
python make_snps_files.py [gff file] [vcf file] [scaf_number] [distance] [out directory] [all/noncoding] [annotation summary file]
You need PyVCF to run this script.
Distance refers to the distance from the TSS or TES that you want to include in the file.
With the all option, you get all SNPs included in the genotype file and with the noncoding option you only get noncoding SNPs If you use the noncoding option, you need to add an annotation summary file. If not, you can use a "." here. You can read more about the annotation format here
After running this you will end up with a bunch of files in your out directory that are named [gene name].[scaffold].snps
For example: 20889218.scaf1.snps
The first line of this file will have the gene name and all the individuals. The subsequent lines will have the SNP coordinate and then genotypes at this coordinate.
They should look like this:
20889218 16A 11G 138Q 203A 114I 24F
1577995 het het het hom1 hom1 hom1
1578031 hom2 hom2 het het het hom2
1578033 hom1 hom1 hom1 hom1 hom1 hom1
note that hom1 is the reference homozygote
This script takes in vcfs for RNA and DNA and outputs a table of allele-specific expression values (ie allelic imbalances) for all genes and all individuals using the RNA vcf information at informative heterozygous sites (genotype is inferred from the DNA vcf)
python aseValuesByGene.py [DNA vcf] [RNA vcf] [out file] [min snps per gene] [list of median expression levels]
min snps per gene is an integer. If set to >1, then it only returns and ASE value if there are more than the specified number of informative heterozygous snps in the gene.
the list of median expression levels is used to normalize the output by sequencing depths. These are median expression levels for all genes from a certain individual. This file should have two columns -- the first column is individual name and the second is median expression.
The output is a table where each row is a gene and each column is an individual and the values are allele-specific expression.
This combined script will run both eQTL and aseQTL analysis on a set of snp files (generated above).
Basic usage is:
python allqtl.py -e [EXP_FILE] -a [ASE_FILE] -s [SNP DIRECTORY] -s [OUTPUT FILE]
If you would like to permute the data randomly
python allqtl.py -e [EXP_FILE] -a [ASE_FILE] -s [SNP DIRECTORY] -s [OUTPUT FILE] -p -n [PERMUTATION NUMBER]
To subsample so that analyses are run on samples of 50 individuals (10 and 40 in each category).
python allqtl.py -e [EXP_FILE] -a [ASE_FILE] -s [SNP DIRECTORY] -s [OUTPUT FILE] -x
To see all options:
python allqtls.py -h
The output is a table with information for each snp. The columns are
scaf: scaffold name
pac: gene name
locus: SNP location
N.eqtl: The number of individuals that could be tested for eQTLs
N.ase: The number of individuals that could be tested for aseQTL
af1: Frequency of the reference allele
maf: Frequency of the minor allele
ase_hom: Mean ASE of homozygotes for the SNP
ase_het: Mean ase for heterozygotes for the snp
h.ase: Mann-Witney statistic for aseQTL test
p.ase: P value for aseQTL test
hom1_mean: Mean expression of individuals homozygous for reference allele
het_mean: Mean expression of individuals heterozygous at SNP
hom2_mean: Mean expression of individuals homozygous for alternate allele
h.eqtl: Mann-Whitney statistic for eQTL test
p.eqtl: P value for eQTL test