molecular QTLs#

molQTL summary statistics#

Options#

--eqtl-flist flistfile#
read flist eQTL format
--add-gene-n geneSampleSizeFile#
Add specific sample size for each gene-snp pair
--keep-gene#
extract genes from files.

Convertion between SMR BESD format and gz format#

Convert plist format to besd format

BayesOmics64 --eqtl-flist flistFile \
    --make-besd --out mybesd

Since BESD format can save effect size and se only, a compressed gz format has been designed to incorporate more useful information

psuedo-1kg-chr22-causal-model.q.query.gz

GeneID             GeneChr  GeneLength   GenePhyPos   SNPID     SNPChr   SNPPhyPos  A1  A2  A1Freq    BETA        SE   Pvalue N
ENSG00000070413       22     2000000     19066881.0   rs2073776   22    19024651    C   T     0.499   0.0015  0.0044  0.72    100000
ENSG00000070413       22     2000000     19066881.0   rs6623      22    19025381    C   T     0.499   0.0016  0.0044  0.70    100000
ENSG00000070413       22     2000000     19066881.0   rs10160     22    19026025    G   A     0.498   -0.003  0.0044  0.44    100000
......

So we can convert BESD format to query.gz format if additional gene sample size is needed

BayesOmics64 --beqtl-summary mybesd \
    --make-query  --out myquery

Please see SMR flist to besd for details.

Extract a single genes basd on gene input file

BayesOmics64 --beqtl-summary mybesd \
    --keep-one-gene geneID \
    --make-query --out myquery

Add gene sample size

There are two ways to add sample size into analysis:

  1. when all SNPs within the same gene have the same sample size,

GeneID        N
ENSG00000182902       10000
ENSG00000183628       10000
ENSG00000100346       10000
......

we suggest that adding sample size information into BESD format directly by the folloing command:

BayesOmics64 --beqtl-summary mybesd \
    --add-gene-n geneSampleSizeFile\
    --make-query \
    --out myquery

  1. when SNPs with the same gene may have different sample size (usually occur in meta analysis, like eQTLs from eQTLGen project),

GeneID        N
ENSG00000182902       10000
ENSG00000183628       10000
ENSG00000100346       10000
......

we suggest that adding these sample size into gz format by the following command:

BayesOmics64 --beqtl-summary mybesd \
    --add-gene-n geneSampleSizeFile\
    --make-query \
    --out myquery

Tip

A) If you’re adding gene sample size to the analysis, the original sample size in the BESD or gz file will be replaced and gene-snp pairs with NA sample size will be deleted automatically. B) If per gene sample size is missing, you can also add total sample size to besd format file by using SMR software, and we will use total sample size for each gene. C) If both total and per gene sample size are missing, we will use estimated gene sample size in the analysis (add this function later).

Warning

Since the LD matrix is the low-rank LD, it’s important to ensure that the gene and SNP set in the geneSampleSizeFile are identical to those in the BESD or gz file. If they’re not identical, you’ll need to perform QC first to ensure they match first.

Merge multiple besd file into one*

BayesOmics64 --beqtl-list mybesdListFile.txt --merge-besd --out mergeBesd

Input file format mybesdListFile.txt (full paths can be specified if the besd files are in different directories)

test_1
test_2
test_3
......
test_22

QC for molQTLs#

It will very convenient to do manipulations on molQTL datasets using R, as the following example:

library(tidyverse)
library(data.table)
eqtl = fread(paste0("example.query.gz"))
####################################
cisWin = 1000000
dat = eqtl %>%
 group_by(GeneChr) %>%
 mutate(
 cisStart = pmax(0, GenePhyPos - cisWin),
 cisEnd = GenePhyPos + cisWin
) %>%
filter(GeneChr == SNPChr) %>%  # Ensure SNP and gene are on the same chromosome
filter(SNPPhyPos >= cisStart & SNPPhyPos <= cisEnd) %>%
filter(abs(BETA) > 1e-6) %>%
select(all_of(colnames(eqtl))) %>%
ungroup() %>%  arrange(GeneChr,GenePhyPos, GeneID, SNPPhyPos)
On this page