K-mer association with mixed effects model

[komaltilwani53]

Hi Team,

Appreciate if someone can advise me on this issue:
Previous steps:

1. SNP and COG association with fixed effects model

Pyseer having issue: K-mer association with mixed effects model

Command used:
pyseer --lmm --phenotypes phenotypes.txt --kmers fsm_kmers.txt.gz --similarity phylogeny_K.tsv --output-patterns kmer_patterns.txt --cpu 12 > cdi_kmers.txt

Standard output: None

Standard Error file:
Read 602 phenotypes
Detected binary phenotype
Setting up LMM
Similarity matrix has dimension (602, 602)
Analysing 602 samples found in both phenotype and similarity matrix
h^2 = 0.00
No observations of TTTNNNNNN in selected samples
No observations of TTTNNNNNNN in selected samples
No observations of TTTNNNNNNNN in selected samples
No observations of TTTNNNNNNNNN in selected samples
No observations of TTTNNNNNNNNNN in selected samples
No observations of TTTNNNNNNNNNNN in selected samples
No observations of TTTNNNNNNNNNNNN in selected samples

Environment Verified and Test cases executed as shared in tutorial.

Please let me know in case if any further inputs required to investigate this issue

[johnlees]

Sorry I'm not clear from the above what is the issue you are having?

[komaltilwani53]

Hi Team,

Appreciate if someone can advise me on this :

These include the result files; the heritibitly score is o, but the Q Q plot is siginificant. This contradicts itself or shouldn't be taken into account for more analysis.
why the heritibility score is 0

1. SNP and COG association with fixed effects model
   Read 602 phenotypes
   Detected binary phenotype
   Structure matrix has dimension (602, 602)
   Analysing 602 samples found in both phenotype and structure matrix
   4701 loaded variants
   2902 pre-filtered variants
   1799 tested variants
   1799 printed variants

2. K-mer association with mixed effects model
   Read 602 phenotypesDetected binary phenotypeSetting up LMMSimilarity matrix has dimension (602, 602)
   Analysing 602 samples found in both phenotype and similarity matrix h^2 = 0.00
   91704889 loaded variants
   5125486 pre-filtered variants
   86579403 tested variants
   86579403 printed variant
   (Patterns: 83884146
   Threshold: 5.96E-10 )

Q-Q plot

[mgalardini]

If I understand correctly, you are asking why you are seeing an heritability estimate of 0, but a number of significant unitigs? Depending on the distribution of your phenotype that is entirely possible (also because it's binary, I think). I don't know your dataset and so this is a little more than guessing

[komaltilwani53]

Please find attached the phenotypic file I have been using. I would be grateful if you could review it and let me know what you think. Do I need to take the findings from this file into account for my analysis?  heritibility score of 0 have a substantial cause
q-q plot is significant

I've been attempting to use Pyseer to create a Manhattan plot from the snp.plot GWAS output. But I haven't been able to locate any noteworthy peaks.

In addition, I would appreciate it if you could provide any techniques or approaches that would enhance my analysis and enable me to get favorable outcomes.

GWAS_Analysis.zip

[mgalardini]

It seems to me that the ratio between positive (1) and negative (0) phenotypes is quite unbalanced (~30 / ~600), which might be a problem. also, from the manhattan plot that you sent I don't see any variant passing the 1E-10 threshold, so maybe you could pick a reference in which the threshold passing variants map to?

Other than that I don't have any particular suggestion

[johnlees]

I wouldn't read too much into h^2 from pyseer, especially with the phenotype as described, it may be heavily biased. You should use another tool if you want to estimate it more accurately