Mu.pp.intersect_obs()

Help muon
1

Hello,

I’m trying to integrate scATAC-seq and scRNA-seq from a 10x Genomics Multiome experiment, but I’m having problems when I run mu.pp.intersect_obs(mdata).

This is the error message:

ValueError: Value passed for key ‘X_pca’ is of incorrect shape. Values of obsm must match dimensions (‘obs’,) of parent. Value had shape (10333,) while it should have had (8988,).

I’m seeing the same error when I try to run mu.pp.filter_obs() in the rna modality.

Thanks,

Leandro

2

Hey @Drito,

Great timing! This has been due to the recent changes in anndata, and we have just fixed that on our end. Do you think you can check the latest version from the github repo and let us know if it works for you? It didn’t make it to a release just yet.

4

Thanks for your reply.
I tried it with the new version and I’m still having the same problem.
Best,

Leandro

5

Hey @Drito,

Thanks for reporting that! Do you think you could provide a minimal example so that we can reproduce that?

6

Hello,

This is a minimal example of my analysis pipeline:

mdata = mu.read_10x_h5(os.path.join(data_dir, "filtered_feature_bc_matrix.h5"))
mdata.var_names_make_unique()

rna = mdata.mod['rna']
atac = mdata.mod['atac']

# RNA
rna.var['mt'] = rna.var_names.str.startswith("MT-") 
sc.pp.calculate_qc_metrics(rna, qc_vars=['mt'], percent_top=None, log1p=False, inplace=True)

mu.pp.filter_var(rna, 'n_cells_by_counts', lambda x: x >= 10)
mu.pp.filter_obs(rna, 'n_genes_by_counts', lambda x: (x >= 500) & (x < 5000))
mu.pp.filter_obs(rna, 'total_counts', lambda x: x < 15000)
mu.pp.filter_obs(rna, 'pct_counts_mt', lambda x: x < 20)

sc.pp.scrublet(rna)
doublets = sum(rna.obs.predicted_doublet)
mu.pp.filter_obs(rna, 'predicted_doublet', lambda x: x==False)
rna.layers['raw_counts'] = rna.X

sc.pp.normalize_total(rna, target_sum=1e4) #1e6 to do CPM normalization
sc.pp.log1p(rna)
rna.layers['lognorm_counts'] = rna.X

sc.pp.highly_variable_genes(rna, min_mean=0.02, max_mean=4, min_disp=0.5)
sc.pp.scale(rna, max_value=10)
rna.layers['scaled_counts'] = rna.X

rna.X = rna.layers['raw_counts'] # Restore original raw counts before Pearson residual normalization
analytic_pearson = sc.experimental.pp.normalize_pearson_residuals(rna, inplace=False)
rna.layers['analytic_pearson_residuals'] = csr_matrix(analytic_pearson['X'])

# ATAC
sc.pp.scrublet(atac)

doublets = sum(atac.obs.predicted_doublet)
mu.pp.filter_obs(atac, 'predicted_doublet', lambda x: x==False)

sc.pp.calculate_qc_metrics(atac, percent_top=None, log1p=False, inplace=True)

# Rename columns to make them fit to scATAC-seq (scanpy labels the metrics according to RNA-seq)
atac.obs.rename(
    columns={
        'n_genes_by_counts': 'n_features_per_cell',
        'total_counts': 'total_fragment_counts'
    },
    inplace=True
)
atac.obs['log_total_fragment_counts'] = np.log10(atac.obs['total_fragment_counts'])

ac.tl.nucleosome_signal(atac)
nuc_signal_threshold = 2
atac.obs['nuc_signal_filter'] = [
    'NS_FAIL' if ns > nuc_signal_threshold else 'NS_PASS'
    for ns in atac.obs['nucleosome_signal']
]

ac.tl.get_gene_annotation_from_rna(mdata['rna']).head(3)
tss = ac.tl.tss_enrichment(mdata, n_tss=2000) 
tss_threshold = 1.5
tss.obs["tss_filter"] = [
    "TSS_FAIL" if score < tss_threshold else "TSS_PASS"
    for score in atac.obs["tss_score"]
]

mu.pp.filter_obs(
    atac,
    "total_fragment_counts",
    lambda x: (x >= 5000) & (x <= 45000),
)
mu.pp.filter_obs(
    atac, 
    "n_features_per_cell", 
    lambda x: x >= 2200
)
mu.pp.filter_obs(
    atac,
    "tss_score",
    lambda x: (x >= 2.0) & (x <= 20),
)
mu.pp.filter_obs(
    atac, 
    "nucleosome_signal", 
    lambda x: x <= 2
)
mu.pp.filter_var(atac, "n_cells_by_counts", lambda x: x >= 10)

atac.layers["raw_counts"] = atac.X 
ac.pp.tfidf(atac, scale_factor=1e4)
atac.layers['tf-idf_norm_counts'] = atac.X

atac.X = atac.layers["raw_counts"] 
sc.pp.normalize_per_cell(atac, counts_per_cell_after=1e4)
sc.pp.log1p(atac)
atac.layers['lognorm_counts'] = atac.X

atac.X = atac.layers['tf-idf_norm_counts'] 

sc.pp.highly_variable_genes(atac, min_mean=0.05, max_mean=1.5, min_disp=0.5)

mu.pp.intersect_obs(mdata)

Thanks,

Leandro

7

Thanks, @Drito!

I can run this code on some test data without any issues.

Both data, on which that can be reproduced, and the full error log would be helpful in order to figure it out.

8

Hello,

I reinstalled muon again from GitHub and now everything works perfectly. Thanks for your help,

Drito

1 Like