Question

plink --assoc command gives "NA" for all the analysis (F_A,F_U,A2,CHISQ,P,OR values)

0

Entering edit mode

5.8 years ago

kushagraprasad24 • 0

Click Here To View The Output Data

After a lot of attempts, i am having the same result.

I am using plink command "--assoc" and resultant output file always gives "NA" values for all of the anaysis.

My input files are in .map and .ped format which were generated from a software "TASSEL 5" (hapmap format to plink format).

Please Help.

SNP plink association analysis • 6.9k views

ADD COMMENT • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

I am not able to find any solution for this type of problem in any other posts.

ADD REPLY • link 5.8 years ago by kushagraprasad24 • 0

0

Entering edit mode

Please use ADD COMMENT or ADD REPLY to answer to previous reactions, as such this thread remains logically structured and easy to follow. I have now moved your reaction but as you can see it's not optimal. Adding an answer should only be used for providing a solution to the question asked.

See also How to add images to a Biostars post

ADD REPLY • link 5.8 years ago by WouterDeCoster 47k

0

Entering edit mode

Please show a sample of your data, and also all commands that you have used. It would have been a good idea to have done this when posting your initial question.

ADD REPLY • link 5.8 years ago by Kevin Blighe 88k

0

Entering edit mode

.ped file

.map file

Sorry I Should Have Done It Earlier.

ADD REPLY • link 5.8 years ago by kushagraprasad24 • 0

0

Entering edit mode

Screenshot Of The Plink Command

ADD REPLY • link 5.8 years ago by kushagraprasad24 • 0

0

Entering edit mode

Well, the problem is right there in your PED file: your phenotype data is full of missing values (-9), so, how can the association test know what to compare?

To confirm, the first 6 columns in your PED file must be:

Family ID
Individual ID
Paternal ID
Maternal ID
Sex (1=male; 2=female; other=unknown)
Phenotype

To run a basic association test, at minimum Individual ID and Phenotype must exist.

ADD REPLY • link 5.8 years ago by Kevin Blighe 88k

0

Entering edit mode

The New .ped File...

--assoc Result...

It Still Returns NA's.

ADD REPLY • link 5.8 years ago by kushagraprasad24 • 0

0

Entering edit mode

...but, what are your phenotypes now? - you just have numbers like 76, 79, etc. What are you actually aiming to compare in the association test? What is the source of your data?

ADD REPLY • link 5.8 years ago by Kevin Blighe 88k

0

Entering edit mode

These are the various phenotype informations...

Im working on wheat genome snp's, the phenotypes that i put are for a single trait...

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

The recent figure to which you linked (above) is not a PED file. It looks like some covariates file that you may use in a linear/logistic regression with PLINK.

Do you understand to what each column in the PED file relates? Do you understand that, when you run the --assoc, it will look at the 7th column in your PED file (Phenotype) and use those groupings for performing the statistical test? If you only have a few samples in each group, then it is difficult to perform any test. GWAS can contain 1000s of samples.

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

Ok, there are 4 groups in my phenotype. It still doesn't return any results. As you said if there are less samples in a group, plink is not able to perform the statistical tests...whats the alternative for this problem?

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

How many total samples do you have? Even with 4 samples per group, it should still return some p-values for some SNPs.

When you show your data, can you literally just paste it here and then wrap it with the 101 010 button. Post, for example, the first few rows of your PED file, and the first 10 of your MAP file. The screenshots are no good because I cannot then take the data and try myself.

Please also show all commands that you used to create the dataset.

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

This is the .ped file :- https://1drv.ms/u/s!AhjOX4irzvPAon9KV5bbrjSzTvqB

This is the .map file :- https://1drv.ms/u/s!AhjOX4irzvPAowCbTXoC_wjCxVc4

You can download the files and try...

I Used TASSEL5 to obtain the .map and .ped file from hapmap format. The original convterted files(.map and .ped) that tassel gave out had all missing values for the phenotype column.

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

Just checking in...has anyone got any solution for this problem?

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

I doubt that many will respond as you have not made it easy for anyone. We just want to see the commands that you have already used, pasted here so that we can easily copy them, and a small / reproducible dataset (also pasted here). By providing screenshots of your data and asking us to download the entire dataset and process it is not helpful - sorry. We are 100% voluntary here and hardly anyone has the time.

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

This is the .ped data:-

-9  450975  -9  -9  -9  2   A   A   C   C   G   G   C   C   A   A   G   G   T   T   A   A   C   C   C   C   A   A   T   T   A   A   T   T   G   G   A   A   A   A   C   C   G
-9  610288  -9  -9  -9  2   0   0   C   C   G   G   T   T   G   G   G   G   C   C   A   A   T   T   T   T   A   A   T   T   A   A   T   T   G   G   A   A   G   G   T   T   G
-9  41868   -9  -9  -9  3   A   A   C   C   G   G   C   C   A   A   G   G   T   T   A   A   C   C   C   C   A   A   T   T   A   A   T   T   G   G   A   A   A   A   C   C   G
-9  3895    -9  -9  -9  2   A   A   C   C   G   G   T   T   A   A   G   G   T   T   A   A   C   C   C   C   A   A   T   T   A   A   T   T   G   G   A   A   A   A   C   C   G
-9  42893   -9  -9  -9  3   0   0   C   C   G   G   T   T   G   G   G   G   C   C   A   A   T   T   T   T   A   A   T   T   A   A   T   T   G   G   A   A   G   G   T   T   G
-9  601549  -9  -9  -9  3   A   A   C   C   G   G   T   T   A   A   G   G   T   T   A   A   C   C   C   C   A   A   T   T   A   A   T   T   G   G   A   A   A   A   C   C   G
-9  80836   -9  -9  -9  2   A   A   C   C   G   G   C   C   A   A   G   G   T   T   A   A   C   C   C   C   A   A   T   T   A   A   T   T   G   G   A   A   A   A   C   C   G

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

This is the .map data:-

1 BS00033749_51 -9  13
1   BS00100774_51   -9  13
1   BobWhite_c1361_1187 -9  13
1   GENE-0412_338   -9  13
1   IAAV3919    -9  13
1   RAC875_c42700_264   -9  13
1   Tdurum_contig44888_837  -9  13
1   wsnp_Ex_c10657_17376448 -9  13
1   BS00026456_51   -9  16
1   Excalibur_c35316_154    -9  16

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

Command i ran :-

  plink --file input --assoc --out result

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

1

Entering edit mode

Thanks, I will try to take a look later. Out of curiosity, you see your 7th column in the PED file? - what if you re-encode the 2s and 3s as 1s and 2s? If you have more than 1 condition, then you need to use a multiple linear regression. --assoc, I believe, only expects 1s and 2s.

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

ok so do i run the multiple linear regression using --linear ? I have 4 groups "1", "2", "3", "4"

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

With 4 groups, -assoc will not function correctly, or it could be that plink determines that your phenotype is on the continuous scale. You have the ability to look through the PLINK manual in order to decide what you now need to do when one has >2 groups. --linear or --logistic seem logical choices.

What do you think, chrchang523 ?

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

This the --logistic result

This is the --linear result

Both Returned NA's.

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

Hey again. Please try to get into the habit of showing the full commands that you have used. You have to minimise the effort on the part of the person trying to help you. Just a useful tip for the future.

Some further points:

I just looked at your PED file again and note that you have -9 as the family ID. I am not sure that that is permitted. Can you change it to 0?
Also, are your phenotypes just encoded 2 and 3? If so, then change them to the default of 1 and 2 - please read the docs for PED file encoding.
In your MAP file, you have multiple SNPs at the same position. I do not believe this is permitted. Your third column in your MAP file is also -9, but I believe it should be 0 if missing

ADD REPLY • link 5.7 years ago by Kevin Blighe 88k

0

Entering edit mode

These are the commands i used :-

 plink --file file --noweb --logistic --out result

 plink --file file --noweb --linear --out result

Yeah i can change it to 0.
I have 4 groups in my phenotype - 1,2,3,4.
ok il try to change the 3rd column to 0's. I dont know what i can do about having multiple snp's at the same positons...

I ran the commands again with the modified files but the result came out to be the same again.

plink --file file --noweb --linear --out result 
plink --file file --noweb --logistic --out result

ADD REPLY • link 5.7 years ago by kushagraprasad24 • 0

0

Entering edit mode

You can exclude SNPs in the same position just to check if everything else is ok.

Use --allow-no-sex option in or commands because you don't have sex information in or .ped file.

ADD REPLY • link 5.3 years ago by martha ▴ 10

0

Entering edit mode

Does this mean if you have more than 2 phenotypes you can only test 2 at a time, i.e. changing the two you are testing to 1,2?

ADD REPLY • link 3.6 years ago by bjwiley23 ▴ 40