.ipynb .pdf

repository open issue suggest edit

Contents

Differential Methylated Genes - Pairwise

import pandas as pd
import anndata
from ALLCools.mcds import MCDS
from ALLCools.clustering import PairwiseDMG

Parameters

adata_path = '../step_by_step/100kb/adata.with_coords.h5ad'
cluster_col = 'L1'

# change this to the paths to your MCDS files
obs_dim = 'cell'
var_dim = 'geneslop2k'

# DMG
mc_type = 'CHN'
top_n = 1000
adj_p_cutoff = 1e-3
delta_rate_cutoff = 0.3
auroc_cutoff = 0.9
random_state = 0
n_jobs = 30

Load

adata = anndata.read_h5ad(adata_path)

cell_meta = adata.obs.copy()
cell_meta.index.name = obs_dim

gene_meta = pd.read_csv(f'GeneMetadata.csv.gz', index_col=0)

gene_mcds = MCDS.open(f'geneslop2k_frac.mcds', use_obs=cell_meta.index)
gene_mcds

<xarray.MCDS>
Dimensions:              (cell: 16985, geneslop2k: 41871, mc_type: 2)
Coordinates:
  * cell                 (cell) <U10 '10E_M_207' '10E_M_338' ... '9J_M_2969'
  * geneslop2k           (geneslop2k) <U21 'ENSMUSG00000102693.1' ... 'ENSMUS...
    geneslop2k_chrom     (geneslop2k) <U5 dask.array<chunksize=(41871,), meta=np.ndarray>
    geneslop2k_cov_mean  (geneslop2k) float64 dask.array<chunksize=(41871,), meta=np.ndarray>
    geneslop2k_end       (geneslop2k) int64 dask.array<chunksize=(41871,), meta=np.ndarray>
    geneslop2k_start     (geneslop2k) int64 dask.array<chunksize=(41871,), meta=np.ndarray>
  * mc_type              (mc_type) <U3 'CGN' 'CHN'
    strand_type          <U4 'both'
Data variables:
    geneslop2k_da_frac   (cell, geneslop2k, mc_type) float32 dask.array<chunksize=(3397, 2463, 2), meta=np.ndarray>
Attributes:
    obs_dim:  cell
    var_dim:  geneslop2k

xarray.MCDS

• Dimensions:
  • cell: 16985
  • geneslop2k: 41871
  • mc_type: 2
• Coordinates: (8)
  • cell
    (cell)
    <U10
    '10E_M_207' ... '9J_M_2969'

    array(['10E_M_207', '10E_M_338', '10E_M_410', ..., '8J_M_590', '8J_M_1809',
           '9J_M_2969'], dtype='<U10')

  • geneslop2k
    (geneslop2k)
    <U21
    'ENSMUSG00000102693.1' ... 'ENSM...

    array(['ENSMUSG00000102693.1', 'ENSMUSG00000064842.1', 'ENSMUSG00000051951.5',
           ..., 'ENSMUSG00000095134.2', 'ENSMUSG00000096768.8',
           'ENSMUSG00000099871.1'], dtype='<U21')

  • geneslop2k_chrom
    (geneslop2k)
    <U5
    dask.array<chunksize=(41871,), meta=np.ndarray>

    Array Chunk Bytes 817.79 kiB 817.79 kiB Shape (41871,) (41871,) Count 2 Tasks 1 Chunks Type numpy.ndarray

  • geneslop2k_cov_mean
    (geneslop2k)
    float64
    dask.array<chunksize=(41871,), meta=np.ndarray>

    Array Chunk Bytes 327.12 kiB 327.12 kiB Shape (41871,) (41871,) Count 2 Tasks 1 Chunks Type float64 numpy.ndarray

  • geneslop2k_end
    (geneslop2k)
    int64
    dask.array<chunksize=(41871,), meta=np.ndarray>

    Array Chunk Bytes 327.12 kiB 327.12 kiB Shape (41871,) (41871,) Count 2 Tasks 1 Chunks Type int64 numpy.ndarray

  • geneslop2k_start
    (geneslop2k)
    int64
    dask.array<chunksize=(41871,), meta=np.ndarray>

    Array Chunk Bytes 327.12 kiB 327.12 kiB Shape (41871,) (41871,) Count 2 Tasks 1 Chunks Type int64 numpy.ndarray

  • mc_type
    (mc_type)
    <U3
    'CGN' 'CHN'

    array(['CGN', 'CHN'], dtype='<U3')

  • strand_type
    ()
    <U4
    'both'

    array('both', dtype='<U4')

• Data variables: (1)
  • geneslop2k_da_frac
    (cell, geneslop2k, mc_type)
    float32
    dask.array<chunksize=(3397, 2463, 2), meta=np.ndarray>

    Array Chunk Bytes 5.30 GiB 63.83 MiB Shape (16985, 41871, 2) (3397, 2463, 2) Count 86 Tasks 85 Chunks Type float32 numpy.ndarray

• Attributes: (2)

  obs_dim :

  cell

  var_dim :

  geneslop2k

Pairwise DMG

pwdmg = PairwiseDMG(max_cell_per_group=1000,
                    top_n=top_n,
                    adj_p_cutoff=adj_p_cutoff,
                    delta_rate_cutoff=delta_rate_cutoff,
                    auroc_cutoff=auroc_cutoff,
                    random_state=random_state,
                    n_jobs=n_jobs)

pwdmg.fit_predict(x=gene_mcds[f'{var_dim}_da_frac'].sel(mc_type=mc_type), 
                  var_dim=var_dim,
                  groups=cell_meta[cluster_col])

Generating cluster AnnData files
Computing pairwise DMG
820 pairwise DMGs

... storing 'groups' as categorical
... storing 'groups' as categorical

pwdmg.dmg_table.to_hdf(f'{cluster_col}.PairwiseDMG.{mc_type}.hdf', key='data')
pwdmg.dmg_table.head()

Aggregating Cluster DMG

Weighted total AUROC aggregated from the pairwise comparisons.

Aggregate Pairwise Comparisons

cluster_dmgs = pwdmg.aggregate_pairwise_dmg(adata, groupby=cluster_col)

# save all the DMGs
with pd.HDFStore(f'{cluster_col}.ClusterRankedPWDMG.{mc_type}.hdf') as hdf:
    for cluster, dmgs in cluster_dmgs.items():
        hdf[cluster] = dmgs[dmgs > 0.0001]

previous

Plot Cluster DMGs

next

Data Integration