Manhattan and Q-Q plot

Load gwaslab

Example

import gwaslab as gl

Load data into Sumstats Object

Example

mysumstats = gl.Sumstats("../0_sample_data/t2d_bbj.txt.gz",
             snpid="SNP",
             chrom="CHR",
             pos="POS",
             ea="ALT",
             nea="REF",
             neaf="Frq",
             p="P",
             build="19",             
             verbose=False)
mysumstats.fix_chr(verbose=False)

SNPID	CHR	POS	EA	NEA	STATUS	EAF	P
1:725932_G_A	1	725932	G	A	1995999	0.9960	0.5970
1:725933_A_G	1	725933	G	A	1995999	0.0040	0.5973
1:737801_T_C	1	737801	C	T	1995999	0.0051	0.6908
1:749963_T_TAA	1	749963	TAA	T	1995999	0.8374	0.2846
1:751343_T_A	1	751343	T	A	1995999	0.8593	0.2705
...	...	...	...	...	...	...	...
X:154874837_A_G	23	154874837	G	A	1995999	0.7478	0.5840
X:154875192_GTACTC_G	23	154875192	GTACTC	G	1995999	0.2525	0.5612
X:154879115_A_G	23	154879115	G	A	1995999	0.7463	0.5646
X:154880669_T_A	23	154880669	T	A	1995999	0.2558	0.5618
X:154880917_C_T	23	154880917	C	T	1995999	0.2558	0.5570

Create Manhattan plot and QQ plot

Example

mysumstats.plot_mqq(skip=2)

stdout:

2025/12/25 23:38:18 Configured plot style for plot_mqq:mqq
2025/12/25 23:38:18 Starting Manhattan-QQ plot creation (Version v4.0.0)
2025/12/25 23:38:18  -Genomic coordinates are based on GRCh37/hg19...
2025/12/25 23:38:18  - Genomic coordinates version: 19 ...
2025/12/25 23:38:18  - Genome-wide significance level to plot is set to 5e-08 ...
2025/12/25 23:38:18  - Input sumstats contains 12557761 variants...
2025/12/25 23:38:18  - Manhattan-QQ plot layout mode selected: mqq
2025/12/25 23:38:19  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:38:20  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:38:20 Finished loading specified columns from the statistics
2025/12/25 23:38:20 Start data conversion and sanity check:
2025/12/25 23:38:25  -Sumstats P values are being converted to -log10(P)...
2025/12/25 23:38:26  -Converting data above cut line...
2025/12/25 23:38:26  -Maximum -log10(P) value is 167.58838029403677 .
2025/12/25 23:38:26 Finished data conversion and sanity check.
2025/12/25 23:38:26 Start to create Manhattan-QQ plot with 332882 variants...
2025/12/25 23:38:26  -Creating background plot...
2025/12/25 23:38:27 Finished creating Manhattan-QQ plot successfully
2025/12/25 23:38:27 Start to extract variants for annotation...
2025/12/25 23:38:27  -Found 89 significant variants with a sliding window size of 500 kb...
2025/12/25 23:38:27 Finished extracting variants for annotation...
2025/12/25 23:38:27 Start to create QQ plot with 332882 variants:
2025/12/25 23:38:27  -Plotting all variants...
2025/12/25 23:38:27  -Expected range of P: (0,1.0)
2025/12/25 23:38:28  -Lambda GC (MLOG10P mode) at 0.5 is   1.21283
2025/12/25 23:38:28 Finished creating QQ plot successfully!
2025/12/25 23:38:28 Finished creating plot successfully

Output image

gl.plot_mqq() Options

Layout mode

If plotting all variants, it may take several minutes. You can use skip to skip variants with low MLOG10P in the plot.

Note: use verbose=False to stop printing log and use check=False to skip sanity check for mqq plots

4 patterns of layout:

mode= "mqq" (default)
mode= "qqm"
mode= "qq"
mode= "m"

Example

mysumstats.plot_mqq(skip=3,cut=20,mode="qqm",check=False, verbose=False)

stdout:

2025/12/25 23:38:32  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:38:34  -CHR data type is already numeric. Skipping conversion...

Output image

Example

mysumstats.plot_mqq(skip=3,cut=20,mode="qqm",check=False, verbose=False)

stdout:

2025/12/25 23:38:40  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:38:42  -CHR data type is already numeric. Skipping conversion...

Output image

Example

mysumstats.plot_mqq(skip=3,cut=20,mode="m",check=False,verbose=False)

stdout:

2025/12/25 23:38:48  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:38:49  -CHR data type is already numeric. Skipping conversion...

Output image

Example

mysumstats.plot_mqq(mode="qq", fig_kwargs= {"figsize":(1,1)},check=False,verbose=False)

Output image

Y axis

skip

skip : skip the variants with low -log10(P) values for plotting

Example

mysumstats.plot_mqq(skip=5,mode= "m", check=False ,verbose=False)

stdout:

2025/12/25 23:39:30  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:39:32  -CHR data type is already numeric. Skipping conversion...

Output image

cut

cut : scale down the -log10(P) for variants above a certain threshold

Example

mysumstats.plot_mqq(cut=20 ,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:39:36  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:39:38  -CHR data type is already numeric. Skipping conversion...

Output image

Make the Y axis jagged to indicate that it has been rescale.

jagged

Example

mysumstats.plot_mqq(skip=3,cut=20, jagged=True,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:06  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:07  -CHR data type is already numeric. Skipping conversion...

Output image

X axis

use_rank

use_rank: if True, GWASLab will use position rank instead of the physical base-pair positions for x aixs.

There will be no gap if use_rank = True

Example

mysumstats.plot_mqq(skip=3,cut=20,use_rank=True,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:12  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:13  -CHR data type is already numeric. Skipping conversion...

Output image

xtight

xtight=True can be used to remove the padding

Example

mysumstats.plot_mqq(xtight=True, skip=3,cut=20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:18  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:19  -CHR data type is already numeric. Skipping conversion...

Output image

chrpad

chrpad: adjust space between each chromosome by max(POS) * chrpad

Example

mysumstats.plot_mqq(chrpad=0.2,xtight=True, skip=3,cut=20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:23  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:25  -CHR data type is already numeric. Skipping conversion...

Output image

Annotation

anno=True

anno=True : annotate all lead variants with chr:pos

Example

mysumstats.plot_mqq(skip=3,cut=20,anno=True,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:29  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:31  -CHR data type is already numeric. Skipping conversion...

Output image

Since there are a large number of novel loci, if we annotate all loci, it will be too messy. Let's only annotate the loci with P<1e-20 by specifying anno_sig_level=1e-20.

Example

mysumstats.plot_mqq(skip=3,cut=20,anno=True, anno_sig_level=1e-40,mode= "m",check=False,verbose=True)

stdout:

2025/12/25 23:41:36 Configured plot style for plot_mqq:m
2025/12/25 23:41:36 Starting Manhattan plot creation (Version v4.0.0)
2025/12/25 23:41:36  -Genomic coordinates are based on GRCh37/hg19...
2025/12/25 23:41:36  - Genomic coordinates version: 19 ...
2025/12/25 23:41:36  - Genome-wide significance level to plot is set to 5e-08 ...
2025/12/25 23:41:36  - Input sumstats contains 12557761 variants...
2025/12/25 23:41:36  - Manhattan plot layout mode selected: m
2025/12/25 23:41:36  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:38  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:41:38 Finished loading specified columns from the statistics
2025/12/25 23:41:38 Start data conversion and sanity check:
2025/12/25 23:41:38  -Sanity check will be skipped.
2025/12/25 23:41:39  -Sumstats P values are being converted to -log10(P)...
2025/12/25 23:41:41  -Converting data above cut line...
2025/12/25 23:41:41  -Maximum -log10(P) value is 167.58838029403677 .
2025/12/25 23:41:41  -Minus log10(P) values above 20 will be shrunk with a shrinkage factor of 10...
2025/12/25 23:41:41 Finished data conversion and sanity check.
2025/12/25 23:41:41 Start to create Manhattan plot with 91234 variants...
2025/12/25 23:41:41  -Creating background plot...
2025/12/25 23:41:41 Finished creating Manhattan plot successfully
2025/12/25 23:41:41 Start to extract variants for annotation...
2025/12/25 23:41:41  -Found 7 significant variants with a sliding window size of 500 kb...
2025/12/25 23:41:41 Finished extracting variants for annotation...
2025/12/25 23:41:41  -Annotating using column CHR:POS...
2025/12/25 23:41:41  -Adjusting text positions with repel_force=0.03...
2025/12/25 23:41:41 Finished creating plot successfully

Output image

anno="GENENAME"

anno="GENENAME" : automatically annotate the nearest gene name

Note: remember to set build=19 or build=38 when loading or plotting.

Example

mysumstats.plot_mqq(skip=3,cut=20,anno="GENENAME", anno_sig_level=1e-20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:42  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:44  -CHR data type is already numeric. Skipping conversion...

Output image

You can specify anno_gtf_path to use your own GTF file for GENENAME annotation

Example

# mysumstats.plot_mqq(anno_gtf_path="/home/yunye/.gwaslab/Homo_sapiens.GRCh37.87.chr.gtf.gz",anno="GENENAME")

anno_d

We can use anno_d to slightly adjust the arrows.

anno_d accepts a dictionary of index of annotation: left/right

For example, 1:"left" means to adjust towards left.

Example

mysumstats.plot_mqq(anno_d={1:"left",2:"right"}, skip=3,cut=20,anno=True, anno_sig_level=1e-20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:53  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:41:54  -CHR data type is already numeric. Skipping conversion...

Output image

anno_scale

We can also use arm_scale to adjust where to put the annotation texts.

For example, arm_scale=1.2 means the default length will be multiplied by a factor of 1.2.

Example

mysumstats.plot_mqq(arm_scale=1.2, skip=3,cut=20,anno=True, anno_sig_level=1e-20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:41:59  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:42:01  -CHR data type is already numeric. Skipping conversion...

Output image

arm_scale_d accepts a dictionary of index of annotation: arm_scale

For example, 1:1.2 means to adjust the arm of the second by a factor of 1.2.

Example

mysumstats.plot_mqq(arm_scale_d={0:0.8,
                                 1:0.7,
                                 2:0.6,
                                 3:0.8}, 
                    skip=3,cut=20,anno=True, anno_sig_level=1e-20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:42:07  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:42:08  -CHR data type is already numeric. Skipping conversion...

Output image

anno_style

GWASLab provides three types of different annotation styles

anno_style="right", anno_style="expand", and anno_style="tight"

Example

mysumstats.plot_mqq(skip=3,cut=20,anno=True,anno_style="expand", anno_sig_level=1e-20,mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:42:14  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:42:15  -CHR data type is already numeric. Skipping conversion...

Output image

Example

mysumstats.plot_mqq(skip=3,cut=20,anno=True,anno_style="tight", anno_sig_level=1e-20,mode= "m", check=False,verbose=False)

stdout:

2025/12/25 23:42:21  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:42:22  -CHR data type is already numeric. Skipping conversion...

Output image

anno_set

If we want to annotate only a subset of variants, we can pass a list of variant IDs to anno_set.

Let's check all lead variants and select only two to annotate.

Example

mysumstats.get_lead(verbose=False).sort_values(by="P")

SNPID	CHR	POS	EA	NEA	STATUS	EAF	P
11:2858546_C_T	11	2858546	C	T	1995999	0.6209	2.580000e-168
9:22132729_A_G	9	22132729	G	A	1995999	0.4367	9.848000e-88
6:20688121_T_A	6	20688121	T	A	1995999	0.5758	2.062000e-85
7:127253550_C_T	7	127253550	C	T	1995999	0.9081	4.101000e-74
X:152908887_G_A	23	152908887	G	A	1995999	0.6792	9.197000e-58
...	...	...	...	...	...	...	...
X:21569920_A_G	23	21569920	G	A	1995999	0.3190	2.616000e-08
6:7226959_C_T	6	7226959	C	T	1995999	0.6657	2.849000e-08
15:90393949_C_CT	15	90393949	C	CT	1995999	0.3445	3.134000e-08
1:154309595_TA_T	1	154309595	TA	T	1995999	0.0947	3.289000e-08
17:40913366_C_T	17	40913366	C	T	1995999	0.4707	4.159000e-08

[89 rows x 8 columns]

This time, let's annotate 1:154309595_TA_T and 2:27734972_G_A with their nearest gene names!

anno_set : the set of variants you want to annotate

Example

mysumstats.plot_mqq(skip=3,cut=20,anno="GENENAME",anno_set=["1:154309595_TA_T","2:27734972_G_A"],mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:42:55  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:42:56  -CHR data type is already numeric. Skipping conversion...

Output image

anno_alias

anno_alias accepts a dictionary of SNPID:string. You can use this to customize the text for annotation.

Example

mysumstats.plot_mqq(skip=3,
                    cut=20,
                    anno=True,
                    anno_set=["1:154309595_TA_T","2:27734972_G_A"],
                    anno_alias={"1:154309595_TA_T":"anything you want here"},
                    mode= "m",
                    verbose=False)

stdout:

2025/12/25 23:43:03  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:04  -CHR data type is already numeric. Skipping conversion...

Output image

Highlight loci & Pinpoint variants (single group)

highlight: a variant list of loci you want to highlight
pinpoint: a variant list of variants you want to pinpoint

Example

mysumstats.plot_mqq(skip=3,cut=20,
                    anno=True,
                    anno_set=["2:27734972_G_A","2:27734972_G_A","7:127253550_C_T", "19:46166604_C_T"],
                    highlight=["19:46166604_C_T","1:154309595_TA_T","7:127253550_C_T"],
                    highlight_windowkb=1000,
                    pinpoint=["2:27734972_G_A"],
                    mode= "m",
                    check=False,verbose=False)

stdout:

2025/12/25 23:43:11  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:13  -CHR data type is already numeric. Skipping conversion...

Output image

Highlight loci & Pinpoint variants (multi-group)

Instead of a list, you can provide a list of lists. Each member list is then a group.

Example

mysumstats.plot_mqq(skip=3,cut=20,
                    highlight=[
                               ["19:46166604_C_T","1:154309595_TA_T"],
                               ["X:57170781_A_AT","7:127253550_C_T"]
                              ],
                    highlight_windowkb=1000,
                    highlight_color=["yellow","red"],
                    mode= "m",
                    check=False,verbose=False)

stdout:

2025/12/25 23:43:20  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:22  -CHR data type is already numeric. Skipping conversion...

Output image

MAF-stratified QQ plot

Example

mysumstats.plot_mqq(skip=3,cut=20, mode="mqq",stratified=True,check=False,verbose=False)

stdout:

2025/12/25 23:43:29  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:30  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:43:30  -EAF data type is already numeric. Skipping conversion...

Output image

Auxiliary lines

Example

mysumstats.plot_mqq(skip=3,
                build="19",
                anno="GENENAME",
                windowsizekb=1000000,
                cut=20,
                cut_line_color="purple",
                sig_level=5e-8,  
                anno_sig_level=1e-6, 
                sig_line_color="grey",
                suggestive_sig_line = True,
                suggestive_sig_level = 1e-6,
                suggestive_sig_line_color="blue",
                additional_line=[1e-40,1e-60],
                additional_line_color=["yellow","green"],
                mode= "m",check=False,verbose=False)

stdout:

2025/12/25 23:43:38  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:40  -CHR data type is already numeric. Skipping conversion...

Output image

Font and marker size

fontsize
anno_fontsize
title_fontsize
marker_size

Example

mysumstats.plot_mqq(skip=3,
          anno="GENENAME",
          title= "My Manhattan Plot",
          windowsizekb=1000000,
          fontsize =13,
          anno_fontsize = 15,
          title_fontsize = 30,
          marker_size=(5,25),
          mode= "m", 
          verbose=False,
          check=False
)

stdout:

2025/12/25 23:43:47  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:48  -CHR data type is already numeric. Skipping conversion...

Output image

Colors

colors
cut_line_color
sig_line_color
highlight_color
pinpoint_color
maf_bin_colors

Example

mysumstats.plot_mqq(skip=3,
          cut=20,
          stratified=True,
          highlight=["7:127253550_C_T"],
          pinpoint=["2:27734972_G_A"],
          colors=["orange","blue"],
          cut_line_color="yellow",
          sig_line_color="red",
          highlight_color="purple",
          pinpoint_color ="green",
          maf_bin_colors = ["#FFE2D1","#E1F0C4", "#6BAB90","#55917F"],
          check=False,verbose=False
)

stdout:

2025/12/25 23:43:55  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:43:56  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:43:56  -EAF data type is already numeric. Skipping conversion...

Output image

Save plots

Example

mysumstats.plot_mqq(skip=3,
                    cut=20, 
                    mode="mqq",
                    stratified=True,
                    save="my_maf_stratified_mqq_plot.png",
                    save_kwargs={"dpi":300,"facecolor":"white"})

stdout:

2025/12/25 23:44:06 Configured plot style for plot_mqq:mqq
2025/12/25 23:44:06 Starting Manhattan-QQ plot creation (Version v4.0.0)
2025/12/25 23:44:06  -Genomic coordinates are based on GRCh37/hg19...
2025/12/25 23:44:06  - Genomic coordinates version: 19 ...
2025/12/25 23:44:06  - Genome-wide significance level to plot is set to 5e-08 ...
2025/12/25 23:44:06  - Input sumstats contains 12557761 variants...
2025/12/25 23:44:06  - Manhattan-QQ plot layout mode selected: mqq
2025/12/25 23:44:06  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:44:08  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:44:08  -EAF data type is already numeric. Skipping conversion...
2025/12/25 23:44:08 Finished loading specified columns from the statistics
2025/12/25 23:44:08 Start data conversion and sanity check:
2025/12/25 23:44:13  -Removed 0 variants with nan in EAF column ...
2025/12/25 23:44:14  -Sumstats P values are being converted to -log10(P)...
2025/12/25 23:44:16  -Converting data above cut line...
2025/12/25 23:44:16  -Maximum -log10(P) value is 167.58838029403677 .
2025/12/25 23:44:16  -Minus log10(P) values above 20 will be shrunk with a shrinkage factor of 10...
2025/12/25 23:44:16 Finished data conversion and sanity check.
2025/12/25 23:44:16 Start to create Manhattan-QQ plot with 91234 variants...
2025/12/25 23:44:16  -Creating background plot...
2025/12/25 23:44:17 Finished creating Manhattan-QQ plot successfully
2025/12/25 23:44:17 Start to extract variants for annotation...
2025/12/25 23:44:17  -Found 89 significant variants with a sliding window size of 500 kb...
2025/12/25 23:44:17 Finished extracting variants for annotation...
2025/12/25 23:44:17 Start to create QQ plot with 91234 variants:
2025/12/25 23:44:17  -Plotting variants stratified by MAF...
2025/12/25 23:44:18  -Lambda GC (MLOG10P mode) at 0.5 is   1.21283
2025/12/25 23:44:18 Finished creating QQ plot successfully!
2025/12/25 23:44:18 Start to save figure...
2025/12/25 23:44:20  -Saved to my_maf_stratified_mqq_plot.png successfully! (png) (overwrite)
2025/12/25 23:44:20 Finished saving figure...
2025/12/25 23:44:20 Finished creating plot successfully

Output image

Fig object

plot_mqq will return a matplotlib figure object

Example

my_mqqplot = mysumstats.plot_mqq(skip=3,cut=20, mode="mqq",stratified=True,verbose=False)

stdout:

2025/12/25 23:44:41  -POS data type is already numeric. Skipping conversion...
2025/12/25 23:44:42  -CHR data type is already numeric. Skipping conversion...
2025/12/25 23:44:42  -EAF data type is already numeric. Skipping conversion...

Example

my_mqqplot

Output image