GWAS pipeline 3 — Spatially Resolved GWAS Mapping

GWAS pipeline 3 — Spatially Resolved GWAS Mapping#

This notebook walks through the complete OmicVerse gsMap integration using the official gsMap example dataset:

find_latent_representation
latent_to_gene
quick-mode generate_ldscore
quick-mode spatial_ldsc
cauchy_combination
Visualisation with the standalone plotting methods

The main dataset is the official gsMap example E16.5_E1S1.MOSTA.h5ad (121,767 spots) and the GWAS trait is IQ_NG_2018.sumstats.gz. All intermediate and final outputs are written to ./gsmap_tutorial_output.

Note: This tutorial uses ov.pl.embedding(basis='spatial') for spatial plotting, which only requires obsm['spatial'] and works without H&E images.

import warnings
warnings.filterwarnings("ignore", category=UserWarning)
warnings.filterwarnings("ignore", category=FutureWarning)

from pathlib import Path
import numpy as np
import pandas as pd
import scanpy as sc
import omicverse as ov
import matplotlib.pyplot as plt

ov.plot_set()
np.random.seed(0)

🔬 Starting plot initialization...
🧬 Detecting GPU devices…
✅ NVIDIA CUDA GPUs detected: 8
    • [CUDA 0] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 1] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 2] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 3] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 4] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 5] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 6] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9
    • [CUDA 7] NVIDIA GeForce RTX 4090 D
      Memory: 23.5 GB | Compute: 8.9

   ____            _     _    __                  
  / __ \____ ___  (_)___| |  / /__  _____________ 
 / / / / __ `__ \/ / ___/ | / / _ \/ ___/ ___/ _ \ 
/ /_/ / / / / / / / /__ | |/ /  __/ /  (__  )  __/ 
\____/_/ /_/ /_/_/\___/ |___/\___/_/  /____/\___/                                              

🔖 Version: 2.2.1rc1   📚 Tutorials: https://omicverse.readthedocs.io/
✅ plot_set complete.

Before you start#

Data used in this tutorial#

ST data: ./gsMap_example_data/ST/E16.5_E1S1.MOSTA.h5ad
GWAS data: ./gsMap_example_data/GWAS/IQ_NG_2018.sumstats.gz
gsMap quick-mode resources: ./gsMap_resource

Output directory#

./gsmap_tutorial_output

Running notes#

The full E16.5_E1S1.MOSTA.h5ad contains more than 100,000 spots. The notebook loads the full official dataset without downsampling.

Step 0 — Load the official gsMap example data#

Read the official MOSTA embryo section and prepare the input for gsMap.

Load E16.5_E1S1.MOSTA.h5ad
Use the built-in annotation column for tissue labels
Preserve the original spatial coordinates in obsm['spatial']
Preserve the count layer for the latent-representation step

# Download the resources:
wget https://yanglab.westlake.edu.cn/data/gsMap/gsMap_resource.tar.gz
tar -xvzf gsMap_resource.tar.gz

# Download example data
wget https://yanglab.westlake.edu.cn/data/gsMap/gsMap_example_data.tar.gz
tar -xvzf gsMap_example_data.tar.gz

dataset_path = Path('./gsMap_example_data/ST/E16.5_E1S1.MOSTA.h5ad')
sumstats_file = Path('./gsMap_example_data/GWAS/IQ_NG_2018.sumstats.gz')
gsmap_resource_dir = Path('./gsMap_resource')
tutorial_workdir = Path('./gsmap_tutorial_output')
adata = sc.read_h5ad(dataset_path)
adata.obs['annotation'] = adata.obs['annotation'].astype('category')
adata

AnnData object with n_obs × n_vars = 121767 × 28204
    obs: 'n_genes_by_counts', 'log1p_n_genes_by_counts', 'total_counts', 'log1p_total_counts', 'annotation', 'Regulon - AI987944', 'Regulon - Alx1', 'Regulon - Alx4', 'Regulon - Arid3a', 'Regulon - Arnt2', 'Regulon - Arx', 'Regulon - Atf1', 'Regulon - Atf2', 'Regulon - Atf3', 'Regulon - Atf4', 'Regulon - Atf6', 'Regulon - Atoh1', 'Regulon - Bach2', 'Regulon - Barhl1', 'Regulon - Barhl2', 'Regulon - Barx1', 'Regulon - Bcl6', 'Regulon - Bclaf1', 'Regulon - Bdp1', 'Regulon - Bhlhe40', 'Regulon - Bmyc', 'Regulon - Brca1', 'Regulon - Brf1', 'Regulon - Brf2', 'Regulon - Cdx1', 'Regulon - Cdx2', 'Regulon - Cebpa', 'Regulon - Cebpb', 'Regulon - Cebpe', 'Regulon - Cic', 'Regulon - Clock', 'Regulon - Creb1', 'Regulon - Creb3l1', 'Regulon - Creb3l2', 'Regulon - Creb5', 'Regulon - Crebl2', 'Regulon - Ctcf', 'Regulon - Ctcfl', 'Regulon - Cux1', 'Regulon - Dbp', 'Regulon - Dbx1', 'Regulon - Ddit3', 'Regulon - Dlx1', 'Regulon - Dlx2', 'Regulon - Dlx3', 'Regulon - Dlx5', 'Regulon - Dlx6', 'Regulon - Dmbx1', 'Regulon - E2f1', 'Regulon - E2f2', 'Regulon - E2f3', 'Regulon - E2f4', 'Regulon - E2f5', 'Regulon - E2f6', 'Regulon - E2f7', 'Regulon - E2f8', 'Regulon - Ebf1', 'Regulon - Egr1', 'Regulon - Egr3', 'Regulon - Elf1', 'Regulon - Elf2', 'Regulon - Elf3', 'Regulon - Elf4', 'Regulon - Elk3', 'Regulon - Elk4', 'Regulon - Emx1', 'Regulon - Emx2', 'Regulon - En1', 'Regulon - En2', 'Regulon - Ep300', 'Regulon - Erg', 'Regulon - Esrra', 'Regulon - Esrrg', 'Regulon - Ets1', 'Regulon - Ets2', 'Regulon - Etv6', 'Regulon - Evx1', 'Regulon - Ezh2', 'Regulon - Fli1', 'Regulon - Fos', 'Regulon - Fosl1', 'Regulon - Fosl2', 'Regulon - Foxa1', 'Regulon - Foxa2', 'Regulon - Foxa3', 'Regulon - Foxc1', 'Regulon - Foxc2', 'Regulon - Foxd1', 'Regulon - Foxd2', 'Regulon - Foxd3', 'Regulon - Foxf1', 'Regulon - Foxf2', 'Regulon - Foxg1', 'Regulon - Foxi1', 'Regulon - Foxj1', 'Regulon - Foxj2', 'Regulon - Foxk1', 'Regulon - Foxl1', 'Regulon - Foxl2', 'Regulon - Foxn2', 'Regulon - Foxn3', 'Regulon - Foxo1', 'Regulon - Foxo3', 'Regulon - Foxo4', 'Regulon - Foxp1', 'Regulon - Foxp2', 'Regulon - Foxp3', 'Regulon - Foxp4', 'Regulon - Foxq1', 'Regulon - Gabpa', 'Regulon - Gabpb1', 'Regulon - Gata1', 'Regulon - Gata2', 'Regulon - Gata3', 'Regulon - Gata4', 'Regulon - Gata5', 'Regulon - Gata6', 'Regulon - Gbx1', 'Regulon - Gbx2', 'Regulon - Gli1', 'Regulon - Gli2', 'Regulon - Glis2', 'Regulon - Gm12845', 'Regulon - Gm14327', 'Regulon - Gm28308', 'Regulon - Gm38394', 'Regulon - Gmeb2', 'Regulon - Grhl1', 'Regulon - Grhl2', 'Regulon - Grhl3', 'Regulon - Gsx1', 'Regulon - Gsx2', 'Regulon - Gtf2f1', 'Regulon - Gtf3c2', 'Regulon - Hcfc1', 'Regulon - Hdac2', 'Regulon - Hes1', 'Regulon - Hes5', 'Regulon - Hif1a', 'Regulon - Hinfp', 'Regulon - Hmga2', 'Regulon - Hmgb3', 'Regulon - Hmgn3', 'Regulon - Hmx1', 'Regulon - Hmx3', 'Regulon - Hnf1b', 'Regulon - Hnf4a', 'Regulon - Hnf4g', 'Regulon - Hoxa10', 'Regulon - Hoxa11', 'Regulon - Hoxa2', 'Regulon - Hoxa3', 'Regulon - Hoxa4', 'Regulon - Hoxa5', 'Regulon - Hoxa7', 'Regulon - Hoxa9', 'Regulon - Hoxb13', 'Regulon - Hoxb3', 'Regulon - Hoxb4', 'Regulon - Hoxb5', 'Regulon - Hoxb6', 'Regulon - Hoxb7', 'Regulon - Hoxb8', 'Regulon - Hoxb9', 'Regulon - Hoxc10', 'Regulon - Hoxc11', 'Regulon - Hoxc13', 'Regulon - Hoxc4', 'Regulon - Hoxc5', 'Regulon - Hoxc6', 'Regulon - Hoxc8', 'Regulon - Hoxc9', 'Regulon - Hoxd1', 'Regulon - Hoxd10', 'Regulon - Hoxd11', 'Regulon - Hoxd12', 'Regulon - Hoxd13', 'Regulon - Hoxd3', 'Regulon - Hoxd8', 'Regulon - Hoxd9', 'Regulon - Hsf2', 'Regulon - Ikzf1', 'Regulon - Irf2', 'Regulon - Irf5', 'Regulon - Irf6', 'Regulon - Irf8', 'Regulon - Irx3', 'Regulon - Isl1', 'Regulon - Isl2', 'Regulon - Jdp2', 'Regulon - Jun', 'Regulon - Junb', 'Regulon - Jund', 'Regulon - Kdm4a', 'Regulon - Kdm5a', 'Regulon - Kdm5b', 'Regulon - Klf10', 'Regulon - Klf13', 'Regulon - Klf15', 'Regulon - Klf16', 'Regulon - Klf2', 'Regulon - Klf3', 'Regulon - Klf4', 'Regulon - Klf5', 'Regulon - Klf6', 'Regulon - Klf7', 'Regulon - Klf8', 'Regulon - Klf9', 'Regulon - Lbx1', 'Regulon - Lef1', 'Regulon - Lhx1', 'Regulon - Lhx2', 'Regulon - Lhx5', 'Regulon - Lhx6', 'Regulon - Lhx8', 'Regulon - Lhx9', 'Regulon - Lmo2', 'Regulon - Lmx1a', 'Regulon - Lmx1b', 'Regulon - Ltf', 'Regulon - Maf', 'Regulon - Mafa', 'Regulon - Mafb', 'Regulon - Mafk', 'Regulon - Max', 'Regulon - Maz', 'Regulon - Mbd1', 'Regulon - Mef2a', 'Regulon - Mef2c', 'Regulon - Mef2d', 'Regulon - Meis1', 'Regulon - Meis2', 'Regulon - Meis3', 'Regulon - Mnx1', 'Regulon - Msc', 'Regulon - Msx1', 'Regulon - Msx2', 'Regulon - Msx3', 'Regulon - Mxi1', 'Regulon - Myb', 'Regulon - Mybl1', 'Regulon - Mybl2', 'Regulon - Myc', 'Regulon - Myf5', 'Regulon - Myf6', 'Regulon - Mynn', 'Regulon - Myod1', 'Regulon - Myog', 'Regulon - Nanos1', 'Regulon - Nelfe', 'Regulon - Neurod1', 'Regulon - Nfatc1', 'Regulon - Nfatc2', 'Regulon - Nfatc3', 'Regulon - Nfatc4', 'Regulon - Nfe2', 'Regulon - Nfe2l1', 'Regulon - Nfe2l2', 'Regulon - Nfic', 'Regulon - Nfil3', 'Regulon - Nfkb1', 'Regulon - Nfkb2', 'Regulon - Nfyb', 'Regulon - Nfyc', 'Regulon - Nhlh2', 'Regulon - Nkx1-1', 'Regulon - Nkx2-1', 'Regulon - Nkx2-4', 'Regulon - Nkx2-5', 'Regulon - Nkx6-1', 'Regulon - Nkx6-2', 'Regulon - Nkx6-3', 'Regulon - Nobox', 'Regulon - Npdc1', 'Regulon - Nr1h3', 'Regulon - Nr1h4', 'Regulon - Nr1i2', 'Regulon - Nr2c1', 'Regulon - Nr2c2', 'Regulon - Nr2f2', 'Regulon - Nr3c1', 'Regulon - Nr4a1', 'Regulon - Nr5a1', 'Regulon - Nr5a2', 'Regulon - Nrf1', 'Regulon - Nucb1', 'Regulon - Olig1', 'Regulon - Olig3', 'Regulon - Onecut1', 'Regulon - Onecut2', 'Regulon - Onecut3', 'Regulon - Otx2', 'Regulon - Ovol2', 'Regulon - Patz1', 'Regulon - Pax5', 'Regulon - Pax6', 'Regulon - Pbx2', 'Regulon - Pbx3', 'Regulon - Pdx1', 'Regulon - Phf8', 'Regulon - Phox2a', 'Regulon - Phox2b', 'Regulon - Pitx1', 'Regulon - Pitx2', 'Regulon - Pknox1', 'Regulon - Plagl1', 'Regulon - Pml', 'Regulon - Pole3', 'Regulon - Pou2f1', 'Regulon - Pou3f1', 'Regulon - Pou3f2', 'Regulon - Pou3f3', 'Regulon - Pou3f4', 'Regulon - Pou4f1', 'Regulon - Pou4f2', 'Regulon - Pou4f3', 'Regulon - Pou6f1', 'Regulon - Ppara', 'Regulon - Pparg', 'Regulon - Ppargc1a', 'Regulon - Prdm16', 'Regulon - Prop1', 'Regulon - Prrx2', 'Regulon - Prrxl1', 'Regulon - Psmd12', 'Regulon - Rad21', 'Regulon - Rara', 'Regulon - Rarb', 'Regulon - Rarg', 'Regulon - Rax', 'Regulon - Rbbp5', 'Regulon - Rcor1', 'Regulon - Rela', 'Regulon - Relb', 'Regulon - Rest', 'Regulon - Rfx2', 'Regulon - Rfx3', 'Regulon - Rfx4', 'Regulon - Rfxap', 'Regulon - Rreb1', 'Regulon - Runx1', 'Regulon - Runx3', 'Regulon - Rxra', 'Regulon - Rxrb', 'Regulon - Rxrg', 'Regulon - Sap30', 'Regulon - Scx', 'Regulon - Setdb1', 'Regulon - Shox2', 'Regulon - Sin3a', 'Regulon - Six1', 'Regulon - Six2', 'Regulon - Six4', 'Regulon - Smad1', 'Regulon - Smad5', 'Regulon - Smarca4', 'Regulon - Smarcb1', 'Regulon - Smarcc2', 'Regulon - Smc3', 'Regulon - Snai3', 'Regulon - Sox10', 'Regulon - Sox11', 'Regulon - Sox12', 'Regulon - Sox13', 'Regulon - Sox15', 'Regulon - Sox17', 'Regulon - Sox18', 'Regulon - Sox2', 'Regulon - Sox21', 'Regulon - Sox3', 'Regulon - Sox4', 'Regulon - Sox7', 'Regulon - Sox8', 'Regulon - Sox9', 'Regulon - Sp1', 'Regulon - Sp2', 'Regulon - Sp3', 'Regulon - Sp4', 'Regulon - Sp6', 'Regulon - Sp7', 'Regulon - Sp8', 'Regulon - Sp9', 'Regulon - Spdef', 'Regulon - Spi1', 'Regulon - Srebf1', 'Regulon - Srebf2', 'Regulon - Srf', 'Regulon - Stat1', 'Regulon - Stat3', 'Regulon - Stat6', 'Regulon - Supt20', 'Regulon - Taf1', 'Regulon - Tagln2', 'Regulon - Tal1', 'Regulon - Tbp', 'Regulon - Tbx1', 'Regulon - Tbx15', 'Regulon - Tbx2', 'Regulon - Tbx4', 'Regulon - Tbx5', 'Regulon - Tcf7', 'Regulon - Tcf7l1', 'Regulon - Tcf7l2', 'Regulon - Tead1', 'Regulon - Tead2', 'Regulon - Tead3', 'Regulon - Tead4', 'Regulon - Tef', 'Regulon - Tfap2a', 'Regulon - Tfcp2l1', 'Regulon - Tfdp1', 'Regulon - Tff3', 'Regulon - Thap1', 'Regulon - Thra', 'Regulon - Thrb', 'Regulon - Tlx2', 'Regulon - Tlx3', 'Regulon - Traf4', 'Regulon - Trim28', 'Regulon - Trp53', 'Regulon - Trp63', 'Regulon - Trp73', 'Regulon - Twist1', 'Regulon - Twist2', 'Regulon - Ubtf', 'Regulon - Uncx', 'Regulon - Usf2', 'Regulon - Vax1', 'Regulon - Vezf1', 'Regulon - Vsx2', 'Regulon - Wt1', 'Regulon - Xbp1', 'Regulon - Xrcc4', 'Regulon - Yy1', 'Regulon - Zbtb14', 'Regulon - Zbtb7b', 'Regulon - Zeb1', 'Regulon - Zfhx2', 'Regulon - Zfhx3', 'Regulon - Zfp110', 'Regulon - Zfp112', 'Regulon - Zfp14', 'Regulon - Zfp143', 'Regulon - Zfp146', 'Regulon - Zfp148', 'Regulon - Zfp189', 'Regulon - Zfp266', 'Regulon - Zfp281', 'Regulon - Zfp369', 'Regulon - Zfp467', 'Regulon - Zfp612', 'Regulon - Zfp641', 'Regulon - Zfp647', 'Regulon - Zfp652', 'Regulon - Zfp672', 'Regulon - Zfp708', 'Regulon - Zfp710', 'Regulon - Zfp729b', 'Regulon - Zfp740', 'Regulon - Zfp950', 'Regulon - Zfp974', 'Regulon - Zfp975', 'Regulon - Zfx', 'Regulon - Zic1', 'Regulon - Zic2', 'Regulon - Zic3', 'Regulon - Zic4', 'Regulon - Zmat4', 'Regulon - Zmiz1', 'Module_1', 'Module_2', 'Module_3', 'Module_4', 'Module_5', 'Module_6', 'Module_7', 'Module_8', 'Module_9', 'Module_10', 'Module_11', 'Module_12', 'Module_13', 'Module_14', 'Module_15', 'Module_16', 'Module_17', 'Module_18', 'Module_19', 'Module_20', 'Module_21', 'Module_22', 'Module_23', 'Module_24', 'Module_25', 'Module_26', 'Module_27', 'Module_28', 'Module_29', 'Module_30', 'Module_31', 'Module_32', 'Module_33', 'Module_34', 'Module_35', 'pct_counts_in_top_50_genes', 'pct_counts_in_top_100_genes', 'pct_counts_in_top_200_genes', 'pct_counts_in_top_500_genes'
    var: 'n_cells', 'n_cells_by_counts', 'mean_counts', 'log1p_mean_counts', 'pct_dropout_by_counts', 'total_counts', 'log1p_total_counts', 'Regulon - Alx4', 'Regulon - Arnt2', 'Regulon - Atf1', 'Regulon - Atf3', 'Regulon - Atf4', 'Regulon - Bach2', 'Regulon - Barhl1', 'Regulon - Barhl2', 'Regulon - Barx1', 'Regulon - Bclaf1', 'Regulon - Brca1', 'Regulon - Cdx1', 'Regulon - Cdx2', 'Regulon - Cebpa', 'Regulon - Cebpb', 'Regulon - Cebpe', 'Regulon - Clock', 'Regulon - Creb1', 'Regulon - Creb3l2', 'Regulon - Ctcfl', 'Regulon - Cux1', 'Regulon - Dbp', 'Regulon - Ddit3', 'Regulon - Dlx1', 'Regulon - Dlx2', 'Regulon - Dlx3', 'Regulon - Dlx5', 'Regulon - Dlx6', 'Regulon - Dmbx1', 'Regulon - E2f1', 'Regulon - E2f3', 'Regulon - E2f4', 'Regulon - E2f5', 'Regulon - E2f6', 'Regulon - E2f7', 'Regulon - E2f8', 'Regulon - Egr1', 'Regulon - Egr3', 'Regulon - Elf2', 'Regulon - Elf4', 'Regulon - Elk3', 'Regulon - Elk4', 'Regulon - Emx1', 'Regulon - Emx2', 'Regulon - En2', 'Regulon - Ep300', 'Regulon - Esrra', 'Regulon - Esrrg', 'Regulon - Ets1', 'Regulon - Ets2', 'Regulon - Etv6', 'Regulon - Ezh2', 'Regulon - Fli1', 'Regulon - Fos', 'Regulon - Fosl2', 'Regulon - Foxa1', 'Regulon - Foxa2', 'Regulon - Foxa3', 'Regulon - Foxc1', 'Regulon - Foxc2', 'Regulon - Foxf1', 'Regulon - Foxf2', 'Regulon - Foxi1', 'Regulon - Foxj1', 'Regulon - Foxj2', 'Regulon - Foxl1', 'Regulon - Foxl2', 'Regulon - Foxn2', 'Regulon - Foxo1', 'Regulon - Foxo3', 'Regulon - Foxo4', 'Regulon - Foxp1', 'Regulon - Foxp2', 'Regulon - Foxp4', 'Regulon - Gabpa', 'Regulon - Gata1', 'Regulon - Gata2', 'Regulon - Gata3', 'Regulon - Gata4', 'Regulon - Gata5', 'Regulon - Gata6', 'Regulon - Gbx2', 'Regulon - Gli1', 'Regulon - Gli2', 'Regulon - Glis2', 'Regulon - Gm38394', 'Regulon - Gmeb2', 'Regulon - Grhl1', 'Regulon - Grhl2', 'Regulon - Grhl3', 'Regulon - Gsx1', 'Regulon - Hdac2', 'Regulon - Hes1', 'Regulon - Hinfp', 'Regulon - Hmga2', 'Regulon - Hmgb3', 'Regulon - Hnf1b', 'Regulon - Hnf4a', 'Regulon - Hnf4g', 'Regulon - Hoxa10', 'Regulon - Hoxa11', 'Regulon - Hoxa2', 'Regulon - Hoxa3', 'Regulon - Hoxa5', 'Regulon - Hoxa7', 'Regulon - Hoxa9', 'Regulon - Hoxb3', 'Regulon - Hoxb4', 'Regulon - Hoxb5', 'Regulon - Hoxb6', 'Regulon - Hoxb7', 'Regulon - Hoxb8', 'Regulon - Hoxb9', 'Regulon - Hoxc10', 'Regulon - Hoxc11', 'Regulon - Hoxc13', 'Regulon - Hoxc6', 'Regulon - Hoxc9', 'Regulon - Hoxd11', 'Regulon - Hoxd12', 'Regulon - Hoxd13', 'Regulon - Hoxd9', 'Regulon - Hsf2', 'Regulon - Ikzf1', 'Regulon - Irf2', 'Regulon - Irf5', 'Regulon - Irf6', 'Regulon - Irf8', 'Regulon - Isl1', 'Regulon - Isl2', 'Regulon - Jun', 'Regulon - Junb', 'Regulon - Jund', 'Regulon - Klf2', 'Regulon - Klf3', 'Regulon - Klf4', 'Regulon - Klf5', 'Regulon - Klf6', 'Regulon - Klf8', 'Regulon - Lef1', 'Regulon - Lhx1', 'Regulon - Lhx2', 'Regulon - Lhx5', 'Regulon - Lhx6', 'Regulon - Lhx8', 'Regulon - Lhx9', 'Regulon - Lmx1a', 'Regulon - Ltf', 'Regulon - Maf', 'Regulon - Mafb', 'Regulon - Mafk', 'Regulon - Mbd1', 'Regulon - Mef2a', 'Regulon - Mef2c', 'Regulon - Mef2d', 'Regulon - Mnx1', 'Regulon - Msc', 'Regulon - Mxi1', 'Regulon - Myb', 'Regulon - Mybl1', 'Regulon - Mybl2', 'Regulon - Myc', 'Regulon - Myf5', 'Regulon - Myf6', 'Regulon - Mynn', 'Regulon - Myod1', 'Regulon - Myog', 'Regulon - Nanos1', 'Regulon - Neurod1', 'Regulon - Nfatc1', 'Regulon - Nfatc4', 'Regulon - Nfe2', 'Regulon - Nfic', 'Regulon - Nfkb2', 'Regulon - Nfyb', 'Regulon - Nkx1-1', 'Regulon - Nkx6-2', 'Regulon - Nr1h3', 'Regulon - Nr1i2', 'Regulon - Nr2c1', 'Regulon - Nr4a1', 'Regulon - Nr5a1', 'Regulon - Nr5a2', 'Regulon - Nucb1', 'Regulon - Onecut1', 'Regulon - Onecut2', 'Regulon - Onecut3', 'Regulon - Otx2', 'Regulon - Ovol2', 'Regulon - Pax6', 'Regulon - Pdx1', 'Regulon - Phox2a', 'Regulon - Phox2b', 'Regulon - Pitx2', 'Regulon - Pou2f1', 'Regulon - Pou3f1', 'Regulon - Pou3f3', 'Regulon - Pou4f3', 'Regulon - Pparg', 'Regulon - Prrx2', 'Regulon - Rara', 'Regulon - Rarb', 'Regulon - Rax', 'Regulon - Rela', 'Regulon - Rfx2', 'Regulon - Rfx3', 'Regulon - Rfx4', 'Regulon - Runx1', 'Regulon - Rxra', 'Regulon - Shox2', 'Regulon - Six1', 'Regulon - Six2', 'Regulon - Six4', 'Regulon - Smad1', 'Regulon - Smad5', 'Regulon - Smarcc2', 'Regulon - Snai3', 'Regulon - Sox10', 'Regulon - Sox11', 'Regulon - Sox12', 'Regulon - Sox13', 'Regulon - Sox17', 'Regulon - Sox18', 'Regulon - Sox2', 'Regulon - Sox21', 'Regulon - Sox4', 'Regulon - Sox7', 'Regulon - Sox9', 'Regulon - Sp1', 'Regulon - Sp3', 'Regulon - Sp4', 'Regulon - Sp7', 'Regulon - Spdef', 'Regulon - Spi1', 'Regulon - Srebf2', 'Regulon - Srf', 'Regulon - Stat6', 'Regulon - Supt20', 'Regulon - Taf1', 'Regulon - Tagln2', 'Regulon - Tal1', 'Regulon - Tbp', 'Regulon - Tbx2', 'Regulon - Tbx5', 'Regulon - Tcf7', 'Regulon - Tcf7l1', 'Regulon - Tcf7l2', 'Regulon - Tead1', 'Regulon - Tead2', 'Regulon - Tead3', 'Regulon - Tead4', 'Regulon - Tfap2a', 'Regulon - Tfcp2l1', 'Regulon - Tfdp1', 'Regulon - Tff3', 'Regulon - Thap1', 'Regulon - Tlx2', 'Regulon - Tlx3', 'Regulon - Trp53', 'Regulon - Trp63', 'Regulon - Twist1', 'Regulon - Uncx', 'Regulon - Xbp1', 'Regulon - Yy1', 'Regulon - Zbtb14', 'Regulon - Zfp112', 'Regulon - Zfp14', 'Regulon - Zfp189', 'Regulon - Zfp641', 'Regulon - Zfp672', 'Regulon - Zfp729b', 'Regulon - Zfp740', 'Regulon - Zfp974', 'Regulon - Zfp975', 'Regulon - Zic2', 'Regulon - Zmat4', 'Module_1', 'Module_10', 'Module_11', 'Module_12', 'Module_13', 'Module_14', 'Module_15', 'Module_16', 'Module_17', 'Module_18', 'Module_19', 'Module_2', 'Module_20', 'Module_21', 'Module_22', 'Module_23', 'Module_24', 'Module_25', 'Module_26', 'Module_27', 'Module_28', 'Module_29', 'Module_3', 'Module_30', 'Module_31', 'Module_32', 'Module_33', 'Module_34', 'Module_35', 'Module_4', 'Module_5', 'Module_6', 'Module_7', 'Module_8', 'Module_9'
    uns: 'annotation_colors'
    obsm: 'spatial'
    varm: 'PCs'
    layers: 'count'

Step 1 — Create the OmicVerse gsMap object#

ov.genetics.gsmap(...) is the public entry point exposed by OmicVerse.

gsmap_object = ov.genetics.gsmap(adata,    
                                 workdir=str(tutorial_workdir),
                                 sample_name='e16_5_e1s1_mosta_full',
                                 annotation='annotation',)

Step 2 — Run `find_latent_representation`#

This step learns a latent embedding from the expression matrix and the spatial structure.

Key parameters (official defaults)#

Parameter	Description	Default
`data_layer='count'`	Which layer to use as input	`'count'`
`epochs=300`	Full training epochs	`300`
`feat_cell=3000`	Number of highly variable genes	`3000`
`feat_hidden1=256, feat_hidden2=128`	Feature encoder hidden dims	`256`, `128`
`gat_hidden1=64, gat_hidden2=30`	GAT hidden dims	`64`, `30`
`n_comps=300`	PCA dimensions	`300`
`n_neighbors=11`	Spatial neighbourhood size	`11`
`nheads=3`	GAT attention heads	`3`

latent_path = gsmap_object.find_latent_representation(
    data_layer='count',
    epochs=300,
    feat_cell=3000,
    feat_hidden1=256,
    feat_hidden2=128,
    gat_hidden1=64,
    gat_hidden2=30,
    n_comps=300,
    n_neighbors=11,
    nheads=3,
)
latent_path

🔍 [2026-05-28 14:24:14] Begin find_latent_representation
🚀 Using GPU for computations.
🔍 Loading ST data of e16_5_e1s1_mosta_full...
    The ST data contains 121767 cells, 28204 genes.
🔍 Preprocessing data...
    Using data layer: count
Begin highly variable gene selection
🔍 Calculating spatial graph...
    The graph contains 1217670 edges, 121767 cells.
    10.00 neighbors per cell on average.
🔍 Finding latent representations for whole ST data...
🔍 Begin GAT-AE model training
✅ Convergence reached at epoch 215/300 (loss=1.4445). Training stopped.
✅ Adding latent representations...
✅ Saving ST data to /data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/find_latent_representations/e16_5_e1s1_mosta_full_add_latent.h5ad
    Time: 140.39 seconds.
✅ find_latent_representation completed successfully.

PosixPath('/data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/find_latent_representations/e16_5_e1s1_mosta_full_add_latent.h5ad')

Step 3 — Run `latent_to_gene`#

latent_to_gene maps the local latent-neighbourhood structure back to gene-level specificity scores (GSS).

Human ST data: If your input is already human gene symbols, omit species and homolog_file. No homolog conversion is needed.

Non-human ST data (e.g. mouse, rat, zebrafish): You must provide both species and homolog_file so that gene names can be mapped to human o rthologs before GWAS enrichment.

Key parameters (official defaults)#

Parameter	Description	Default
`num_neighbour=51`	Latent-space neighbours per spot	`51`
`num_neighbour_spatial=201`	Spatial-coordinate neighbours per spot	`201`

marker_path = gsmap_object.latent_to_gene(
    input_hdf5_path=str(latent_path),
    latent_representation='latent_GVAE',
    num_neighbour=51,
    num_neighbour_spatial=201,
    species='MOUSE_GENE_SYM',
    homolog_file='./gsMap_resource/homologs/mouse_human_homologs.txt',
)
marker_path

🔍 [2026-05-28 14:26:35] Begin latent_to_gene
🔍 Loading the spatial data...
    Loaded spatial data with 121767 cells and 28204 genes.
🔍 Cell annotations are provided as annotation...
🔍 Transforming MOUSE_GENE_SYM to HUMAN_GENE_SYM...
    16331 genes retained after homolog transformation.
    16331 genes retained after removing duplicates.
🔍 Building the spatial graph...
🔍 Building spatial graph based on spatial coordinates...
    Cell annotations are provided.
    Cavity: 11287 cells
    Epidermis: 6304 cells
    Connective tissue: 8803 cells
    Muscle: 9853 cells
    Adipose tissue: 3822 cells
    Cartilage primordium: 5850 cells
    Submandibular gland: 2007 cells
    Jaw and tooth: 3078 cells
    Bone: 3400 cells
    Cartilage: 5602 cells
    Lung: 3760 cells
    Kidney: 2158 cells
    Meninges: 6656 cells
    GI tract: 1227 cells
    Liver: 14167 cells
    Inner ear: 457 cells
    Adrenal gland: 194 cells
    Dorsal root ganglion: 2047 cells
    Mucosal epithelium: 2783 cells
    Smooth muscle: 383 cells
    Heart: 3723 cells
    Sympathetic nerve: 634 cells
    Spinal cord: 4471 cells
    Brain: 17374 cells
    Choroid plexus: 1727 cells
✅ Spatial graph built successfully.
🔍 Extracting the latent representation...
✅ Latent representation extracted.
🔍 Ranking the spatial data...
✅ Gene expression proportion of each gene across cells computed.
🔍 Computing marker scores...
✅ Marker scores computed.
    Removed mitochondrial genes. Remaining genes: 16331.
🔍 Saving marker scores...
✅ Marker scores saved to /data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/latent_to_gene/e16_5_e1s1_mosta_full_gene_marker_score.feather.
✅ Modified adata object saved to /data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/find_latent_representations/e16_5_e1s1_mosta_full_add_latent.h5ad.
    Time: 1061.80 seconds.
✅ latent_to_gene completed successfully.

PosixPath('/data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/latent_to_gene/e16_5_e1s1_mosta_full_gene_marker_score.feather')

Step 4 — Quick-mode `generate_ldscore`#

Connect pre-computed LD resources (baseline annotations, SNP–gene pairs and LDSC weights) to the marker scores produced above.

ldscore_dir = gsmap_object.generate_ldscore(
    gsmap_resource_dir=str(gsmap_resource_dir),
)
ldscore_dir

🔍 [2026-05-28 14:44:16] Begin generate_ldscore
🔍 Running gsMap generate_ldscore in quick_mode with precomputed resources.
    Time: 0.01 seconds.
✅ generate_ldscore completed successfully.

PosixPath('/data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/generate_ldscore')

Step 5 — Run quick-mode `spatial_ldsc`#

Combine marker scores, LD scores and GWAS summary statistics to obtain a per-spot association p-value.

Note: After this step completes, p-values are automatically written back to the latent adata as {trait}_gsmap_p (raw p) and {trait}_gsmap_logp (-log10(p)) and saved to the latent h5ad file. No manual loading is needed for downstream visualization.

Key parameters (official defaults)#

Parameter	Description	Default
`num_processes=4`	Parallel worker processes	`4`
`n_blocks=200`	Jackknife block count	`200`
`spots_per_chunk_quick_mode=1000`	Spots processed per chunk	`1000`

ldsc_dir = gsmap_object.spatial_ldsc(
    gsmap_resource_dir=str(gsmap_resource_dir),
    sumstats_file=str(sumstats_file),
    trait_name='IQ',
    num_processes=4,
    n_blocks=200,
    spots_per_chunk_quick_mode=1000,
)

🔍 [2026-05-28 14:44:16] Begin spatial_ldsc
    Time: 2763.43 seconds.
✅ spatial_ldsc completed successfully.

Step 6 — Aggregate spot-level signal with `cauchy_combination`#

Aggregate per-spot p-values into annotation-level p-values using the Cauchy combination test, yielding one combined p-value per tissue region.

cauchy_file = gsmap_object.cauchy_combination(
    trait_name='IQ',
    annotation='annotation',
)
cauchy_file

🔍 [2026-05-28 15:30:20] Begin cauchy_combination
✅ Cauchy combination results saved at /data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/cauchy_combination/e16_5_e1s1_mosta_full_IQ.Cauchy.csv.gz.
    Time: 4.29 seconds.
✅ cauchy_combination completed successfully.

PosixPath('/data/hulei/gsmap_omicverse_tutorial_run/e16_5_e1s1_mosta_full/cauchy_combination/e16_5_e1s1_mosta_full_IQ.Cauchy.csv.gz')

Step 7 — Visualise results with the standalone plotting methods#

The OmicVerse gsMap integration adds five standalone plotting methods that generate matplotlib / plotly figures directly, without waiting for the full HTML report.

7.1 Spatial visualization of gsMap p-values#

Note: plot_gsmap_spatial has been removed. After spatial_ldsc runs, p-values are automatically written back to adata.obs as {trait}_gsmap_p and {trait}_gsmap_logp. Use sc.pl.spatial directly to visualize them.

adata_plot = gsmap_object._get_latent_adata()
fig, ax = plt.subplots(figsize=(3,4))
ov.pl.embedding(adata_plot, basis="spatial", color=['IQ_gsmap_logp'], cmap='YlOrRd', size=1.25,ax=ax)

../_images/43db1d966edb53b04feb122a0635f6bf5f8f8042b00b3950c88363b9c25012ce.png

7.2 `plot_cauchy_bar` — Cauchy-combination p-value bar chart#

Note: Bars are now rendered with a continuous gradient (default YlOrRd), uniform 12 pt fonts, and top/right spines removed. Pass cmap to change the color map.

gsmap_object.plot_cauchy_bar("IQ", cmap="Blues",figsize=(6,8))

../_images/810bced322e2e3de35bb32492c7f02ddaff966b4de02660638f164dbfcc58414.png

7.3 Native `ov.pl.embedding` for gene expression and GSS#

ov.pl.spatial requires adata.uns['spatial'] to contain H&E image metadata, which the MOSTA dataset does not provide. We therefore use ov.pl.embedding(basis='spatial'), which plots directly from adata.obsm['spatial'].

The GSS results are stored in the latent_to_gene directory as a feather file. We load them into adata.obs and call the native OmicVerse API for visualisation.

import pandas as pd
import omicverse as ov
import matplotlib.pyplot as plt

adata_plot = ov.read(gsmap_object.hdf5_with_latent_path)
mkscore_path = gsmap_object.mkscore_feather_path
mk_score = pd.read_feather(mkscore_path).set_index("HUMAN_GENE_SYM").T

gene = 'MAP2'
adata_plot.obs[f"{gene}_expr"] = adata_plot[:, gene].X.toarray().flatten()
adata_plot.obs[f"{gene}_gss"] = mk_score[gene].reindex(adata_plot.obs_names).values

fig, axes = plt.subplots(1, 2, figsize=(6, 4))
ov.pl.embedding(adata_plot, basis='spatial', color=f"{gene}_expr", 
                cmap='YlOrRd', size=1.25, title=f"{gene} Expression", 
                ax=axes[0], show=False)
ov.pl.embedding(adata_plot, basis='spatial', color=f"{gene}_gss", 
                cmap='YlOrRd', size=1.25, title=f"{gene} GSS", 
                ax=axes[1], show=False)

<Axes: title={'center': 'MAP2 GSS'}, xlabel='spatial1', ylabel='spatial2'>

../_images/cea8545e42c7e4e0601f2351e92f6ce1d74bd27cd380900e34625117f11b865e.png

7.4 Batch-plot top-correlated genes (native API)#

Read the top-correlated genes from the Gene Diagnostic Info CSV, write their GSS into adata.obs, and plot them together with ov.pl.embedding.

from pathlib import Path

gene_diag_path = Path(gsmap_object.workdir) / gsmap_object.sample_name / 'report' / 'IQ' / f'{gsmap_object.sample_name}_IQ_Gene_Diagnostic_Info.csv'
gene_info = pd.read_csv(gene_diag_path)
top_genes = gene_info.Gene.iloc[:5].tolist()
print("Top 5 genes:", top_genes)

for g in top_genes:
    adata_plot.obs[f"{g}_gss"] = mk_score[g].reindex(adata_plot.obs_names).values

fig, axes = plt.subplots(2, 3, figsize=(9, 8))
axes = axes.flatten()  

for idx, g in enumerate(top_genes):
    ov.pl.embedding(adata_plot,basis="spatial",color=f"{g}_gss",cmap="YlOrRd",size=1.25,ax=axes[idx],show=False)
    axes[idx].set_title(f"{g} GSS", fontsize=9)

for idx in range(len(top_genes), len(axes)):
    axes[idx].set_visible(False)

Top 5 genes: ['MAP2', 'MYT1L', 'ELAVL3', 'CDK5R1', 'CNIH2']

../_images/769b577222dbc53ac0db10689e9ca93aaa35b86b2af75a71188f5ad8752ab63a.png

Optional: one-shot pipeline#

Run the currently integrated pipeline in a single call:

result = gsmap_object.run_pipeline(
    find_latent_kwargs={'data_layer': 'counts', 'epochs': 300, 'n_comps': 300},
    latent_to_gene_kwargs={'latent_representation': 'latent_GVAE'
                          'species': 'MOUSE_GENE_SYM',
                          'homolog_file': './gsMap_resource/homologs/mouse_human_homologs.txt',},
    generate_ldscore_kwargs={'gsmap_resource_dir': str(gsmap_resource_dir)},
    spatial_ldsc_kwargs={
        'gsmap_resource_dir': str(gsmap_resource_dir),
        'sumstats_file': str(sumstats_file),
        'trait_name': 'IQ',
    },
    cauchy_combination_kwargs={'trait_name': 'IQ', 'annotation': 'annotation'},
)

References#

Please cite the relevant tools and resources when using this workflow:

Song, L., Chen, W., Hou, J. et al. Spatially resolved mapping of cells associated with human complex traits. Nature 641, 932–941 (2025). https://doi.org/10.1038/s41586-025-08757-x
Zeng, Z., Ma, Y., Hu, L. et al. OmicVerse: a framework for bridging and deepening insights across bulk and single-cell sequencing. Nat Commun 15, 5983 (2024). https://doi.org/10.1038/s41467-024-50194-3

GWAS pipeline 3 — Spatially Resolved GWAS Mapping

Contents

GWAS pipeline 3 — Spatially Resolved GWAS Mapping#

Before you start#

Data used in this tutorial#

Output directory#

Running notes#

Step 0 — Load the official gsMap example data#

Step 1 — Create the OmicVerse gsMap object#

Step 2 — Run find_latent_representation#

Key parameters (official defaults)#

Step 3 — Run latent_to_gene#

Key parameters (official defaults)#

Step 4 — Quick-mode generate_ldscore#

Step 5 — Run quick-mode spatial_ldsc#

Key parameters (official defaults)#

Step 6 — Aggregate spot-level signal with cauchy_combination#

Step 7 — Visualise results with the standalone plotting methods#

7.1 Spatial visualization of gsMap p-values#

7.2 plot_cauchy_bar — Cauchy-combination p-value bar chart#

7.3 Native ov.pl.embedding for gene expression and GSS#

7.4 Batch-plot top-correlated genes (native API)#

Optional: one-shot pipeline#

References#

Step 2 — Run `find_latent_representation`#

Step 3 — Run `latent_to_gene`#

Step 4 — Quick-mode `generate_ldscore`#

Step 5 — Run quick-mode `spatial_ldsc`#

Step 6 — Aggregate spot-level signal with `cauchy_combination`#

7.2 `plot_cauchy_bar` — Cauchy-combination p-value bar chart#

7.3 Native `ov.pl.embedding` for gene expression and GSS#