how to get predicted expression level,Tg?

[ymqwa]

Hi,
I have GWAS summary data and individual level data in the same cohort, also have DNA methylation prediction models. I use GWAS summary data to run Predict.py, get the methylation-level association results. Now i want to get the predicted methylation value, just like predicted expression Tg . i dont know if it can be got ? and if can , how to do ?

Thank you,
best mqyang.

[Fnyasimi]

have a look at this for individual level data and this for summary stats.

I hope this helps

[ymqwa]

Really appreciate! @Fnyasimi

but when i run Predict.py, I got the following output:
INFO - Loading samples
INFO - Loading model
INFO - Preparing genotype dosages
INFO - Setting whitelist from available models
INFO - Processing genotypes
INFO - Preparing prediction
INFO - Couldn't import h5py_cache. Anyway, this dependency should be removed. It has been folded into h5py
Level 9 - Processing vcfs
Level 9 - Processing vcf chrall_qc_3.vcf.gz
Traceback (most recent call last):
File "/Public/xyzhao/data/methylation/metaxcan/MetaXcan-master/software/Predict.py", line 270, in
run(args)
File "/Public/xyzhao/data/methylation/metaxcan/MetaXcan-master/software/Predict.py", line 198, in run
results.update(gene, dosage, weight)
AttributeError: 'NoneType' object has no attribute 'update'

Here is my code:
conda activate MetaXcan
/Public/software/MetaXcan/Predict.py
--model_db_path methylation_r01_FHS.db
--vcf_genotypes chrall_qc_3.vcf.gz
--vcf_mode genotyped
--prediction_output methy_predict.txt
--prediction_summary_output methy_predict_summary.txt
--verbosity 9
--throw

i don't know how to fix it. Would you have any thoughts on that?
Looking forward to your reply
best mqyang.

[DoraCChen]

Hi, I also had a similar question -

Since Summary-PrediXcan directly computes the gene-level association results from summary GWAS data, is there any way to derive and output predicted gene expression level Tg? I know that for individual-level PrediXcan it is possible to get Tg output using the Predict.py script, but the association analyses are different between individual and summary-level PrediXcan. How can we extract this information? I looked through the wiki page that was linked for S-PrediXcan but wasn't able to find the relevant information.

[hakyim]

Unfortunately, the individual level information cannot be recovered from the summary statistics. To calculate Tg for a set of individuals, you will need their genotype data. You can use your own cohort's genotype data and apply the individual level predixcan prediction software `predict.py` or use a reference genotype such as 1000G or others. I hope this helps Haky

…

On Mon, Apr 4, 2022 at 1:11 PM Dorothy Chen ***@***.***> wrote: Hi, I also had a similar question - Since Summary-PrediXcan directly computes the gene-level association results from summary GWAS data, is there any way to derive and output predicted gene expression level Tg? I know that for individual-level PrediXcan it is possible to get Tg output using the Predict.py script, but the association analyses are different between individual and summary-level PrediXcan. How can we extract this information? I looked through the wiki page that was linked for S-PrediXcan but wasn't able to find the relevant information. — Reply to this email directly, view it on GitHub <#149 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAW2ROMNQH5J57UOHESQRTTVDMWDZANCNFSM5R27ORWQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>