The SNPs are aligned in a line at the top in the Manhattan plot

[Shaogan Wang]

Hi everyone,

I want to perform GWAS on 239 maize inbred lines using 2.88 million clean-up SNPs in TASSEL5. I downloaded and installed tassel-5-standalone on Google Cloud Platform (Ubuntu 26 server). I run the following command lines:

run_pipeline.pl -Xmx10g \ -fork1 -plink -ped ./maize_tassel_clean.ped -map ./maize_tassel_clean.map \ -fork2 -r ./tassel_phe.txt \ -fork3 -r ./tassel_cov.txt \ -fork4 -kinship -method Centered_IBS -input1 \ -combine5 -input1 -input2 -input3 -input4 -intersect \ -mlm -mlmVarCompEst P3D -export tassel_Sac_gwas_output

When I received the output files, I plotted the data to generate Manhattan and Q-Q plots. I found that both plots were very weird, as shown in the figures. The SNPs are aligned in a line at the top in the Manhattan plot. Could you tell me where the problems are and how to fix them?  I really appreciate your kind help!

 

[Brandon Monier]

Hello,

Are you running any filtration steps before you run MLM to account for MAF and sites found in most of your 239 lines?

Best,

Brandon M.

[Shaogan Wang]

Hello Brandon,

Thanks so much for your question. Actually, the 2.8 million SNPs were filtered from the raw dataset using the following command lines:

plink --allow-extra-chr\
  --biallelic-only strict\
  --double-id\
  --geno 0.05\
  --maf 0.05\
  --make-bed\
  --mind 0.05\
  --out maize239_plink_qc\
  --snps-only\
  --vcf all_chrs_239.vcf.gz
 

plink --bfile maize239_plink_qc \
  --indep-pairwise 50 10 0.2 \
  --out maize239

plink  --bfile maize239_plink_qc\

  --extract maize239.prune.in \
  --make-bed \
  --out maize239_pruned

[Shaogan Wang]

import os
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from bioinfokit import visuz

# 1. Target files
gwas_input_file = "tassel_Sac_gwas_output2.txt"
manhattan_output = "gwas_manhattan_plot"
qq_output = "gwas_qq_plot.png"

print(f"Loading GWAS statistics from: {gwas_input_file}...")
if not os.path.exists(gwas_input_file):
    print(f"Error: Could not find file '{gwas_input_file}' in this folder.")
    exit()

# Load the file
df = pd.read_csv(gwas_input_file, sep="\t")

# Filter out 'None' intercept AND the 1e-9 TASSEL mathematical ceiling
df_filtered = df[(df["Marker"] != "None") & (df["p"] != 1e-9)].copy()

# Enforce numerical formatting
df_filtered["Chr"] = pd.to_numeric(df_filtered["Chr"])
df_filtered["Pos"] = pd.to_numeric(df_filtered["Pos"])
df_filtered["p"] = pd.to_numeric(df_filtered["p"])
df_filtered = df_filtered.dropna(subset=["Marker", "Chr", "Pos", "p"])

print(f"Processed {len(df_filtered)} true SNPs (excluding 1e-9 artifacts).")

# 3. Draw Manhattan Plot using bioinfokit
print("Generating clean Manhattan Plot...")
visuz.marker.mhat(
    df=df_filtered,
    chr="Chr",
    pv="p",
    gwas_sign_line=True,
    gwasp=5e-8,
    figtype="png",
    figname=manhattan_output,
    dim=(12, 6),
    dotsize=5,
    valpha=0.7
)
print(f"Manhattan plot successfully saved: {manhattan_output}.png")

# 4. Generate Q-Q Plot manually (Fixed Sorting Alignment)
print("Generating Q-Q Plot manually...")
n_snps = len(df_filtered)

# Sort observed p-values descending (so -log10 values are ascending: 0 -> max)
p_observed = np.sort(df_filtered["p"].values)[::-1]
log_p_observed = -np.log10(p_observed)

# Compute expected uniform p-values in descending order (so -log10 values are ascending: 0 -> max)
p_expected = np.arange(n_snps, 0, -1) / (n_snps + 1)
log_p_expected = -np.log10(p_expected)

# Render the plot
plt.figure(figsize=(6, 6))
plt.scatter(log_p_expected, log_p_observed, c="crimson", s=5, alpha=0.6, zorder=2)

# Draw the 45-degree diagonal null hypothesis line
max_val = max(max(log_p_expected), max(log_p_observed))
plt.plot([0, max_val], [0, max_val], color="black", linestyle="--", linewidth=1.5, zorder=1)

# Formatting
plt.title("GWAS Quantile-Quantile (Q-Q) Plot", fontsize=12, fontweight="bold")
plt.xlabel(r"Expected $-\log_{10}(P)$", fontsize=10)
plt.ylabel(r"Observed $-\log_{10}(P)$", fontsize=10)
plt.grid(True, which="both", linestyle=":", alpha=0.5)
plt.tight_layout()

# Save the final rendering
plt.savefig(qq_output, dpi=300)
plt.close()
print(f"Q-Q plot successfully saved: {qq_output}")

These are the scripts I used to create Manhattan and Q-Q plots

On Friday, June 12, 2026 at 4:59:23 PM UTC+2 Brandon Monier wrote:

[Atit Parajuli]

How does the PC 1 vs PC2 plot looks like?

--
You received this message because you are subscribed to the Google Groups "TASSEL - Trait Analysis by Association, Evolution and Linkage" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tassel+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/tassel/06a5166e-0067-4a9a-b6a7-5ec1aca7e5c5n%40googlegroups.com.