UMI dedup single panel

We can use UMIche to draw deduplication performance of Markov clustering versus its two important parameters, inflation and expansion.

Tabulate the statistics

Python

from umiche.io import stat

scenarios = {
    # 'pcr_nums': 'PCR cycle',
    # 'pcr_errs': 'PCR error',
    'seq_errs': 'Sequencing error',
    # 'ampl_rates': 'Amplification rate',
    # 'umi_lens': 'UMI length',
    # 'seq_deps': 'Sequencing depth',
}
methods = {
    # 'unique': 'Unique',
    # 'cluster': 'Cluster',
    # 'adjacency': 'Adjacency',
    'directional': 'Directional',
    # 'dbscan_seq_onehot': 'DBSCAN',
    # 'birch_seq_onehot': 'Birch',
    # 'aprop_seq_onehot': 'Affinity Propagation',
    'mcl': 'MCL',
    'mcl_val': 'MCL-val',
    'mcl_ed': 'MCL-ed',
}
dedupstat = stat(
    scenarios=scenarios,
    methods=methods,
    param_fpn=to('data/params.yml'),
)

df_dedup = dedupstat.df_dedup
df_dedup_perm_melt = dedupstat.df_dedup_perm_melt

Define uc.plot.dedup_single for better disease biology study.

Python

import umiche as uc

t = uc.plot.dedup_single(
    df_dedup=df_dedup,
    df_dedup_perm_melt=df_dedup_perm_melt,
)
t.strip()

Image title — **Fig** 1. Jointgrid plot.

Python

t.jointgrid()

Python

t.stackedbar()