Tutorial: Analyzing single cell calling cards collected from the mouse cortex#

In this tutorial, we will analyze single cell calling data that describes Brd4 binding in the mouse cortex. The dataset comes from Moudgil et al., Cell. (2020) and can be downloaded from GEO.

In this tutorial, e will cover how to call peaks, annotatate them, perform a differential peak analysis, and pair peaks with genes. In this dataset, there are 111382 insertions and 35950 cells in total. However, many cells are filtered in scRNA-seq analysis. It uses Mudata for calling cards and RNA data. If you want to use Anndata only, please check Github

[1]:

import pycallingcards as cc
import numpy as np
import pandas as pd
import scanpy as sc
from mudata import MuData
from matplotlib import pyplot as plt
plt.rcParams['figure.dpi'] = 150

We start by reading the qbed datafile. In this file, each row represents a Brd4-directed insertion and columns indicate the chromosome, start point and end point, reads number, the direction and cell barcode of each insertion. For example, the first row indicates one transposition is located on Chromosome 1, and starts from 3112541 and ends on 3112545. The reads number is 12 with the direction going from 3’ to 5’. The barcode of the cell is GATGAAAAGAGTTGGC-1. Note that the barcodes of cells should be consistent with scRNA-seq data.

Use cc.rd.read_qbed(filename) to read your own qbed data.

[2]:

qbed_data = cc.datasets.mousecortex_data(data = "qbed")
qbed_data

[2]:
Chr Start End Reads Direction Barcodes
0 chr1 3112541 3112545 12 + GATGAAAAGAGTTGGC-1
1 chr1 3121337 3121341 6 - CGATCGGCACATTTCT-1
2 chr1 3199281 3199285 7 + GTCCTCATCTCCGGTT-1
3 chr1 3211433 3211437 22 - CGAGAAGAGGAATCGC-1
4 chr1 3245859 3245863 149 + TTTACTGCATCCGCGA-1
... ... ... ... ... ... ...
111377 chrY 90807968 90807972 200 - ACGGAGAGTCGCATAT-1
111378 chrY 90833531 90833535 51 - TAGCCGGTCCTGTACC-1
111379 chrY 90833600 90833604 13 - TTGGCAAAGAATTGTG-1
111380 chrY 90840262 90840266 8 - GTGCATAGTACCAGTT-1
111381 chrY 90840262 90840266 95 + TTGTAGGTCGAATCCA-1

111382 rows × 6 columns

We next need to call peaks in order to find Brd4 binding sites. Three different methods (CCcaller, cc_tools, Blockify) are available along with three different species (hg38, mm10, sacCer3).

In this setting, we will use CCcaller in mouse(‘mm10’) data. maxbetween is the most important parameter for CCcaller. It controls the maximum distance between two nearby insertions, or, in other words, the minimum distance between two peaks. 1000-2000 is a good parameter for maxbetween. pvalue_cutoff is also an important parameter, and a number between 0.001 and 0.05 is strongly advised. pseudocounts is advised to be 0.1-1.

[3]:

peak_data = cc.pp.call_peaks(qbed_data, method = "CCcaller", reference = "mm10",  maxbetween = 2000,
                             pvalue_cutoff = 0.01, lam_win_size = 1000000,  pseudocounts = 1, record = True,
                             save = 'mouse_cortex.bed')
peak_data

For the CCcaller method without background, [expdata, reference, pvalue_cutoff, lam_win_size, pseudocounts, minlen, extend, maxbetween, test_method, min_insertions, record] would be utilized.

100%|██████████| 21/21 [00:16<00:00,  1.24it/s]

[3]:
Chr Start End Experiment Insertions Reference Insertions Expected Insertions pvalue pvalue_adj
0 chr1 4806673 4809049 12 20 1.120541 2.498336e-10 6.563055e-07
1 chr1 14302176 14310895 14 92 1.523252 1.015569e-10 2.845105e-07
2 chr1 15287495 15288141 8 4 1.029800 1.427167e-06 2.002598e-03
3 chr1 18307949 18310271 8 31 1.151983 3.511291e-06 4.395841e-03
4 chr1 18976012 18982286 13 62 1.452142 5.507813e-10 1.357847e-06
... ... ... ... ... ... ... ... ...
696 chrX 158919208 158925514 9 39 1.272189 9.683526e-07 1.413241e-03
697 chrX 165325630 165327490 8 17 1.047701 1.640240e-06 2.254207e-03
698 chrX 166241178 166243587 8 18 1.257844 7.050309e-06 8.151358e-03
699 chrX 166345453 166350005 11 35 1.507674 7.204418e-08 1.263100e-04
700 chrX 169842873 169845831 9 28 1.314399 1.292113e-06 1.842126e-03

701 rows × 8 columns

In order to tune parameters for peak calling, we advise looking at the data and evaluating the validity of the called peaks. The default settings are recommended, but for some TFs, adjacent peaks may be merged that should not be, or, alternatively, peaks that should be joined may be called separately.

In this plot, the top section is insertions and their read counts. One dot is an insertion and the height is log(reads+1). The middle section is the distribution of insertions. The bottom section represents the reference genes and peaks.

[4]:

cc.pl.draw_area("chr12", 50102917, 50124960, 400000, peak_data, qbed_data, "mm10", font_size=2, plotsize = [1,1,6],
                figsize = (30,8), peak_line = 5, save = True, title = "chr12_50102917_50124960")
../../_images/tutorials_notebooks_Mouse_cortex_Example_8_0.png

We can also visualize our data directly in the WashU Epigenome Browser. This can be useful for overlaying your data with other published datasets. Please note that this link only valid for 24hrs, so you will have to rerun it if you want to use it after this time period.

[5]:

qbed = {"qbed_data":qbed_data}
bed = {"peak":peak_data}
cc.pl.WashU_browser_url(qbed = qbed, bed = bed, genome = 'mm10')

All qbed addressed
All bed addressed
Uploading files
Please click the following link to see the data on WashU Epigenome Browser directly.
https://epigenomegateway.wustl.edu/browser/?genome=mm10&hub=https://companion.epigenomegateway.org//task/ce3f3ef6f66e901b7fc00587e885fd07/output//datahub.json

Pycallingcards can be used to visual peak locations acorss the genome to see that the distribution of peaks is unbiased and that all chromosomes are represented.

[6]:

cc.pl.whole_peaks(peak_data, reference = "mm10")
../../_images/tutorials_notebooks_Mouse_cortex_Example_12_0.png

In the next step, we annotate the peaks by their closest genes using bedtools and pybedtools. Make sure these programs are properly installed before using.

[7]:

peak_annotation = cc.pp.annotation(peak_data, reference = "mm10")
peak_annotation = cc.pp.combine_annotation(peak_data, peak_annotation)
peak_annotation

In the bedtools method, we would use bedtools in the default path. Set bedtools path by 'bedtools_path' if needed.

[7]:
Chr Start End Experiment Insertions Reference Insertions Expected Insertions pvalue pvalue_adj Nearest Refseq1 Gene Name1 Direction1 Distance1 Nearest Refseq2 Gene Name2 Direction2 Distance2
0 chr1 4806673 4809049 12 20 1.120541 2.498336e-10 6.563055e-07 NM_008866 Lypla1 + 0 NR_033530 Mrpl15 - -20948
1 chr1 14302176 14310895 14 92 1.523252 1.015569e-10 2.845105e-07 NM_010164 Eya1 - 0 NM_010827 Msc - 442451
2 chr1 15287495 15288141 8 4 1.029800 1.427167e-06 2.002598e-03 NM_001098528 Kcnb2 + 24311 NM_177781 Trpa1 - -368634
3 chr1 18307949 18310271 8 31 1.151983 3.511291e-06 4.395841e-03 NM_183124 Defb41 - -42812 NM_001039123 Defb18 - -70507
4 chr1 18976012 18982286 13 62 1.452142 5.507813e-10 1.357847e-06 NM_153154 Tfap2d + 120736 NM_001286340 Tfap2b + 226628
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
696 chrX 158919208 158925514 9 39 1.272189 9.683526e-07 1.413241e-03 NM_001346675 Rps6ka3 + 330268 NM_025437 Eif1ax + 446681
697 chrX 165325630 165327490 8 17 1.047701 1.640240e-06 2.254207e-03 NM_183427 Glra2 - 0 NM_175027 Fancb + -328359
698 chrX 166241178 166243587 8 18 1.257844 7.050309e-06 8.151358e-03 NM_023122 Gpm6b + 0 NM_001310724 Gemin8 + -50667
699 chrX 166345453 166350005 11 35 1.507674 7.204418e-08 1.263100e-04 NM_023122 Gpm6b + 0 NM_177429 Ofd1 - 40028
700 chrX 169842873 169845831 9 28 1.314399 1.292113e-06 1.842126e-03 NM_010797 Mid1 + 0 NM_001290506 Mid1 + 33788

701 rows × 16 columns

Then, we read the barcode file.

[8]:

barcodes = cc.datasets.mousecortex_data(data = "barcodes")
barcodes

[8]:
index
0 AAACCTGAGAACTCGG-1
1 AAACCTGAGCAATCTC-1
2 AAACCTGAGCCGTCGT-1
3 AAACCTGAGTAGCGGT-1
4 AAACCTGAGTGGAGTC-1
... ...
35945 TTTGTCAAGTCCCACG-1
35946 TTTGTCACAGCGTCCA-1
35947 TTTGTCACATTTCACT-1
35948 TTTGTCAGTCGCATCG-1
35949 TTTGTCATCTTTACAC-1

35950 rows × 1 columns

Use qbed data, peak data and barcode data to make a cell by peak Anndata object.

[9]:

adata_cc = cc.pp.make_Anndata(qbed_data, peak_annotation, barcodes)
adata_cc

100%|██████████| 20/20 [00:00<00:00, 89.37it/s]

[9]:

AnnData object with n_obs × n_vars = 35950 × 701
    var: 'Chr', 'Start', 'End', 'Experiment Insertions', 'Reference Insertions', 'Expected Insertions', 'pvalue', 'pvalue_adj', 'Nearest Refseq1', 'Gene Name1', 'Direction1', 'Distance1', 'Nearest Refseq2', 'Gene Name2', 'Direction2', 'Distance2'

In single cell calling cards, for each cell, RNA-seq data is collected simultaneously with TF binding information. For the following steps, we are going to read scRNA-seq data and analyze them together. Scanpy is recommended to load and analyze scRNA-seq data.

[10]:

adata = cc.datasets.mousecortex_data(data = "RNA")
adata

/home/juanru/miniconda3/lib/python3.9/site-packages/anndata/__init__.py:51: FutureWarning: `anndata.read` is deprecated, use `anndata.read_h5ad` instead. `ad.read` will be removed in mid 2024.
  warnings.warn(

[10]:

AnnData object with n_obs × n_vars = 30300 × 2638
    obs: 'batch', 'n_genes', 'total_counts', 'cluster'
    var: 'n_counts', 'n_cells', 'highly_variable'

In scRNA-seq analysis, many cells are filtered out because of low quality. We need to make the cells in qbed anndata to be the exactly same as RNA-seq anndata.

[11]:

adata_cc = cc.pp.filter_adata_sc(adata_cc, adata)
adata_cc

[11]:

View of AnnData object with n_obs × n_vars = 30300 × 701
    var: 'Chr', 'Start', 'End', 'Experiment Insertions', 'Reference Insertions', 'Expected Insertions', 'pvalue', 'pvalue_adj', 'Nearest Refseq1', 'Gene Name1', 'Direction1', 'Distance1', 'Nearest Refseq2', 'Gene Name2', 'Direction2', 'Distance2'

[12]:

mdata = MuData({"RNA": adata, "CC": adata_cc})
mdata

[12]:

MuData object with n_obs × n_vars = 30300 × 3339
  2 modalities
    RNA:    30300 x 2638
      obs:  'batch', 'n_genes', 'total_counts', 'cluster'
      var:  'n_counts', 'n_cells', 'highly_variable'
    CC:     30300 x 701
      var:  'Chr', 'Start', 'End', 'Experiment Insertions', 'Reference Insertions', 'Expected Insertions', 'pvalue', 'pvalue_adj', 'Nearest Refseq1', 'Gene Name1', 'Direction1', 'Distance1', 'Nearest Refseq2', 'Gene Name2', 'Direction2', 'Distance2'

Next, we will cluster cells by the RNA-seq data to identify cell types so that we can identify differentially bound peaks between different cell types.

Although one peak should have many insertions, but there is a chance that all the cells from the peak were filtered by the RNA preprocesssing. In this case, we advise to filter peaks by the minimum number of cells.

[13]:

cc.pp.filter_peaks(mdata["CC"], min_counts = 1)

/home/juanru/miniconda3/lib/python3.9/site-packages/scanpy/preprocessing/_simple.py:248: ImplicitModificationWarning: Trying to modify attribute `.var` of view, initializing view as actual.
  adata.var['n_counts'] = number

Differential peak analysis shows significant bindings for each cluster. In this example, we use binomial test to find out.

[14]:

cc.tl.rank_peak_groups_mu(mdata, "RNA:cluster", method = 'binomtest', key_added = 'binomtest')
mdata

100%|██████████| 18/18 [00:21<00:00,  1.19s/it]

[14]:

MuData object with n_obs × n_vars = 30300 × 3339
  2 modalities
    RNA:    30300 x 2638
      obs:  'batch', 'n_genes', 'total_counts', 'cluster'
      var:  'n_counts', 'n_cells', 'highly_variable'
    CC:     30300 x 701
      var:  'Chr', 'Start', 'End', 'Experiment Insertions', 'Reference Insertions', 'Expected Insertions', 'pvalue', 'pvalue_adj', 'Nearest Refseq1', 'Gene Name1', 'Direction1', 'Distance1', 'Nearest Refseq2', 'Gene Name2', 'Direction2', 'Distance2', 'n_counts'
      uns:  'binomtest'

Plot the results for differential peak analysis.

Currently, the peaks are ranked by pvalues. It could also be ranked by logfoldchanges by the following code:

cc.tl.rank_peak_groups_mu(mdata, "RNA:cluster", method = 'binomtest', rankby = 'logfoldchanges')
cc.pl.rank_peak_groups(mdata["CC"], key = 'binomtest', rankby = 'logfoldchanges')

[15]:

cc.pl.rank_peak_groups(mdata["CC"], key = 'binomtest', save = True)
../../_images/tutorials_notebooks_Mouse_cortex_Example_30_0.png

Next, we will visualize differentially bound peaks. The colored datapoints are the insertions specific to a cluster and the grey ones are the total insertions across the rest of the clusters. We can see that most of the insertions are from Astrocyte in the following peaks.

[16]:

cc.pl.draw_area_mu("chr3", 34638588, 34656047, 400000, peak_data, qbed_data, "mm10", mdata = mdata, font_size = 2,
                   name = 'Astrocyte', key = 'RNA:cluster', figsize = (30,12), peak_line = 4, color = "blue",
                   name_insertion1 = 'Astrocyte Insertions', name_density1 = 'Astrocyte Insertion Density',
                   name_insertion2 = 'Total Insertions', name_density2 = 'Total Insertion Density',
                   plotsize = [1,1,5], title = "chr3_34638588_34656047")
../../_images/tutorials_notebooks_Mouse_cortex_Example_32_0.png

Next we can ask if differentially bound peaks are near differentially expressed genes, which might suggest the idenified enhancer regulates the nearby gene.

We first perform a differential expression analysis for the scRNA-seq data.

[17]:

sc.tl.rank_genes_groups(mdata["RNA"], 'cluster')

WARNING: Default of the method has been changed to 't-test' from 't-test_overestim_var'

Next, we find peak-gene pairs that are differentially bound/regulated in the specified cell-type. We look at all differential peaks in each cluster and see if the annotated genes are significantly expressed in the cluster. We can then set the pvalue and score/log foldchange cutoff.

[18]:

cc.tl.pair_peak_gene_sc_mu(mdata, pvalue_adj_cutoff_cc = 0.05, pvalue_adj_cutoff_rna = 0.05,
                           lfc_cutoff = 3, score_cutoff = 3)
mdata["CC"].uns['pair']

[18]:
Cluster Peak Logfoldchanges Pvalue_peak Pvalue_adj_peak Gene Score_gene Pvalue_gene Pvalue_adj_gene Distance_peak_to_gene
0 Astrocyte chr16_43501178_43518253 3.077262 1.462410e-16 2.562874e-14 Zbtb20 79.234276 0.000000e+00 0.000000e+00 0
1 Astrocyte chr8_64645834_64659215 4.623955 3.114365e-14 2.949104e-12 Msmo1 61.964111 0.000000e+00 0.000000e+00 58930
2 Astrocyte chr3_141928409_141939733 4.876938 3.786296e-14 2.949104e-12 Bmpr1b 24.131538 2.475603e-121 1.659753e-120 0
3 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 Nfia 45.417362 0.000000e+00 0.000000e+00 188838
4 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 E130114P18Rik 16.893749 3.380841e-62 1.514206e-61 0
... ... ... ... ... ... ... ... ... ... ...
57 Neuron_Excit_L2-4 chr13_83141353_83148478 3.495487 3.141216e-06 2.446659e-04 Mef2c 182.292236 0.000000e+00 0.000000e+00 355556
58 Neuron_Excit_L5_Mixed chr7_66014128_66014532 3.737143 2.786909e-04 2.858012e-02 Pcsk6 -16.587570 2.263189e-61 3.313281e-60 0
59 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ugt8a -23.040115 5.201659e-89 3.379286e-87 454686
60 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ndst4 4.497158 8.449939e-06 3.032315e-05 0
61 Neuron_Granule_DG chr1_50860999_50861403 4.356917 1.071190e-04 1.251507e-02 Tmeff2 3.399986 7.231398e-04 1.952972e-03 66120

62 rows × 10 columns

Plot the results above to find out the potential relationship between TF bindings and gene expression.

[19]:

cc.pl.dotplot_sc_mu(mdata, figsize=(10, 60))
../../_images/tutorials_notebooks_Mouse_cortex_Example_39_0.png

After seeing the dotplot above, bring some peaks to the RNA Anndata object and see the distribution in the UMAP plot.

[20]:

sc.pp.pca(mdata["RNA"])
sc.pp.neighbors(mdata["RNA"])
sc.tl.umap(mdata["RNA"])

The first plot is the average insertions of the peak in each cluster, the second plot is the gene expression Gou3f2 (one of its nearest genes for the peak); the third plot is the cluster information.

[21]:

cc.pl.plot_matched(mdata, 'chr4_22969921_22973019', 'Pou3f2')

/home/juanru/miniconda3/lib/python3.9/site-packages/scanpy/plotting/_tools/scatterplots.py:391: UserWarning: No data for colormapping provided via 'c'. Parameters 'cmap' will be ignored
  cax = scatter(
../../_images/tutorials_notebooks_Mouse_cortex_Example_43_1.png 
[22]:

cc.pl.plot_matched(mdata, 'chr1_42253068_42265192', 'Pou3f3')

The history saving thread hit an unexpected error (OperationalError('database is locked')).History will not be written to the database.

/home/juanru/miniconda3/lib/python3.9/site-packages/scanpy/plotting/_tools/scatterplots.py:391: UserWarning: No data for colormapping provided via 'c'. Parameters 'cmap' will be ignored
  cax = scatter(
../../_images/tutorials_notebooks_Mouse_cortex_Example_44_2.png

We can see a potential relationship between Brd4 binding and gene expression.

If we map these enhancers to the human genome, are there disease associated SNPs nearby? To answer this question, we can map binding sites and genes to the human genome. We use liftover to get the resuls.

[23]:

mdata["CC"].uns["pair"] = cc.tl.result_mapping(mdata["CC"].uns["pair"])
mdata["CC"].uns["pair"]

Start mapping the peaks to the new genome.

100%|██████████| 62/62 [00:00<00:00, 222.90it/s]

Start finding location of genes in the new genome.

100%|██████████| 62/62 [00:00<00:00, 257.11it/s]

[23]:
Cluster Peak Logfoldchanges Pvalue_peak Pvalue_adj_peak Gene Score_gene Pvalue_gene Pvalue_adj_gene Distance_peak_to_gene Chr_liftover Start_liftover End_liftover Chr_hg38 Start_hg38 End_hg38
0 Astrocyte chr16_43501178_43518253 3.077262 1.462410e-16 2.562874e-14 Zbtb20 79.234276 0.000000e+00 0.000000e+00 0 chr3 114439800 114457200 chr3 114314499 115147280
1 Astrocyte chr8_64645834_64659215 4.623955 3.114365e-14 2.949104e-12 Msmo1 61.964111 0.000000e+00 0.000000e+00 58930 chr4 165418445 165438029 chr4 165327665 165343162
2 Astrocyte chr3_141928409_141939733 4.876938 3.786296e-14 2.949104e-12 Bmpr1b 24.131538 2.475603e-121 1.659753e-120 0 chr4 95040417 95054959 chr4 94757976 95158450
3 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 Nfia 45.417362 0.000000e+00 0.000000e+00 188838 chr1 60858298 60872361 chr1 61077273 61462788
4 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 E130114P18Rik 16.893749 3.380841e-62 1.514206e-61 0 chr1 60858298 60872361
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
57 Neuron_Excit_L2-4 chr13_83141353_83148478 3.495487 3.141216e-06 2.446659e-04 Mef2c 182.292236 0.000000e+00 0.000000e+00 355556 chr5 89235156 89249323 chr5 88718240 88904105
58 Neuron_Excit_L5_Mixed chr7_66014128_66014532 3.737143 2.786909e-04 2.858012e-02 Pcsk6 -16.587570 2.263189e-61 3.313281e-60 0 chr15 101337048 101337565 chr15 101303927 101489984
59 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ugt8a -23.040115 5.201659e-89 3.379286e-87 454686 chr4 115106106 115112630
60 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ndst4 4.497158 8.449939e-06 3.032315e-05 0 chr4 115106106 115112630 chr4 114827772 115113876
61 Neuron_Granule_DG chr1_50860999_50861403 4.356917 1.071190e-04 1.251507e-02 Tmeff2 3.399986 7.231398e-04 1.952972e-03 66120 chr2 191949045 192194933

62 rows × 16 columns

We search the GWAS Catalog database and find out related SNPs in the binding areas.

[24]:

mdata["CC"].uns["pair"] = cc.tl.GWAS(mdata["CC"].uns["pair"])
mdata["CC"].uns["pair"]

[24]:
Cluster Peak Logfoldchanges Pvalue_peak Pvalue_adj_peak Gene Score_gene Pvalue_gene Pvalue_adj_gene Distance_peak_to_gene Chr_liftover Start_liftover End_liftover Chr_hg38 Start_hg38 End_hg38 GWAS
0 Astrocyte chr16_43501178_43518253 3.077262 1.462410e-16 2.562874e-14 Zbtb20 79.234276 0.000000e+00 0.000000e+00 0 chr3 114439800 114457200 chr3 114314499 115147280 Schizophrenia; Smoking status (ever vs never s...
1 Astrocyte chr8_64645834_64659215 4.623955 3.114365e-14 2.949104e-12 Msmo1 61.964111 0.000000e+00 0.000000e+00 58930 chr4 165418445 165438029 chr4 165327665 165343162 Atopic dermatitis (moderate to severe)
2 Astrocyte chr3_141928409_141939733 4.876938 3.786296e-14 2.949104e-12 Bmpr1b 24.131538 2.475603e-121 1.659753e-120 0 chr4 95040417 95054959 chr4 94757976 95158450
3 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 Nfia 45.417362 0.000000e+00 0.000000e+00 188838 chr1 60858298 60872361 chr1 61077273 61462788 Refractive error
4 Astrocyte chr4_97575305_97588788 3.099679 5.907086e-14 4.140867e-12 E130114P18Rik 16.893749 3.380841e-62 1.514206e-61 0 chr1 60858298 60872361 Refractive error
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
57 Neuron_Excit_L2-4 chr13_83141353_83148478 3.495487 3.141216e-06 2.446659e-04 Mef2c 182.292236 0.000000e+00 0.000000e+00 355556 chr5 89235156 89249323 chr5 88718240 88904105 Macular thickness; Waist circumference adjuste...
58 Neuron_Excit_L5_Mixed chr7_66014128_66014532 3.737143 2.786909e-04 2.858012e-02 Pcsk6 -16.587570 2.263189e-61 3.313281e-60 0 chr15 101337048 101337565 chr15 101303927 101489984
59 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ugt8a -23.040115 5.201659e-89 3.379286e-87 454686 chr4 115106106 115112630
60 Neuron_Granule_DG chr3_125405334_125410657 5.806064 1.011677e-04 1.251507e-02 Ndst4 4.497158 8.449939e-06 3.032315e-05 0 chr4 115106106 115112630 chr4 114827772 115113876
61 Neuron_Granule_DG chr1_50860999_50861403 4.356917 1.071190e-04 1.251507e-02 Tmeff2 3.399986 7.231398e-04 1.952972e-03 66120 chr2 191949045 192194933

62 rows × 17 columns

Save the file if needed.

[25]:

mdata.write("Mouse-Cortex.h5mu")