Compare outcomes from differential analysis based on different imputation methods

• load scores based on 10_1_ald_diff_analysis

Show code cell source

Hide code cell source

import logging
from pathlib import Path

import matplotlib.pyplot as plt
import pandas as pd
import seaborn as sns
from IPython.display import display

import pimmslearn
import pimmslearn.databases.diseases

logger = pimmslearn.logging.setup_nb_logger()

plt.rcParams['figure.figsize'] = (2, 2)
fontsize = 5
pimmslearn.plotting.make_large_descriptors(fontsize)
logging.getLogger('fontTools').setLevel(logging.ERROR)

# catch passed parameters
args = None
args = dict(globals()).keys()

Parameters

Default and set parameters for the notebook.

folder_experiment = 'runs/appl_ald_data/plasma/proteinGroups'

target = 'kleiner'
model_key = 'VAE'
baseline = 'RSN'
out_folder = 'diff_analysis'
selected_statistics = ['p-unc', '-Log10 pvalue', 'qvalue', 'rejected']

disease_ontology = 5082  # code from https://disease-ontology.org/
# split diseases notebook? Query gene names for proteins in file from uniprot?
annotaitons_gene_col = 'PG.Genes'

# Parameters
disease_ontology = 10652
folder_experiment = "runs/alzheimer_study"
target = "AD"
baseline = "PI"
model_key = "VAE"
out_folder = "diff_analysis"
annotaitons_gene_col = "None"

Add set parameters to configuration

Show code cell source

Hide code cell source

params = pimmslearn.nb.get_params(args, globals=globals())
args = pimmslearn.nb.Config()
args.folder_experiment = Path(params["folder_experiment"])
args = pimmslearn.nb.add_default_paths(args,
                                 out_root=(
                                     args.folder_experiment
                                     / params["out_folder"]
                                     / params["target"]
                                     / f"{params['baseline']}_vs_{params['model_key']}"))
args.update_from_dict(params)
args.scores_folder = scores_folder = (args.folder_experiment
                                      / params["out_folder"]
                                      / params["target"]
                                      / 'scores')
args.freq_features_observed = args.folder_experiment / 'freq_features_observed.csv'
args

root - INFO     Removed from global namespace: folder_experiment
root - INFO     Removed from global namespace: target
root - INFO     Removed from global namespace: model_key
root - INFO     Removed from global namespace: baseline
root - INFO     Removed from global namespace: out_folder
root - INFO     Removed from global namespace: selected_statistics
root - INFO     Removed from global namespace: disease_ontology
root - INFO     Removed from global namespace: annotaitons_gene_col
root - INFO     Already set attribute: folder_experiment has value runs/alzheimer_study
root - INFO     Already set attribute: out_folder has value diff_analysis

{'annotaitons_gene_col': 'None',
 'baseline': 'PI',
 'data': PosixPath('runs/alzheimer_study/data'),
 'disease_ontology': 10652,
 'folder_experiment': PosixPath('runs/alzheimer_study'),
 'freq_features_observed': PosixPath('runs/alzheimer_study/freq_features_observed.csv'),
 'model_key': 'VAE',
 'out_figures': PosixPath('runs/alzheimer_study/figures'),
 'out_folder': PosixPath('runs/alzheimer_study/diff_analysis/AD/PI_vs_VAE'),
 'out_metrics': PosixPath('runs/alzheimer_study'),
 'out_models': PosixPath('runs/alzheimer_study'),
 'out_preds': PosixPath('runs/alzheimer_study/preds'),
 'scores_folder': PosixPath('runs/alzheimer_study/diff_analysis/AD/scores'),
 'selected_statistics': ['p-unc', '-Log10 pvalue', 'qvalue', 'rejected'],
 'target': 'AD'}

Excel file for exports

files_out = dict()
writer_args = dict(float_format='%.3f')

fname = args.out_folder / 'diff_analysis_compare_methods.xlsx'
files_out[fname.name] = fname
writer = pd.ExcelWriter(fname)
logger.info("Writing to excel file: %s", fname)

root - INFO     Writing to excel file: runs/alzheimer_study/diff_analysis/AD/PI_vs_VAE/diff_analysis_compare_methods.xlsx

Load scores

Load baseline model scores

Show all statistics, later use selected statistics

Show code cell source

Hide code cell source

fname = args.scores_folder / f'diff_analysis_scores_{args.baseline}.pkl'
scores_baseline = pd.read_pickle(fname)
scores_baseline

	model	PI
	var	SS	DF	F	p-unc	np2	-Log10 pvalue	qvalue	rejected
protein groups	Source
A0A024QZX5;A0A087X1N8;P35237	AD	0.180	1	0.291	0.590	0.002	0.229	0.728	False
	age	0.084	1	0.135	0.713	0.001	0.147	0.817	False
	Kiel	2.216	1	3.588	0.060	0.018	1.224	0.140	False
	Magdeburg	5.950	1	9.634	0.002	0.048	2.657	0.010	True
	Sweden	10.249	1	16.595	0.000	0.080	4.169	0.000	True
...	...	...	...	...	...	...	...	...	...
S4R3U6	AD	0.367	1	0.390	0.533	0.002	0.273	0.680	False
	age	1.453	1	1.542	0.216	0.008	0.666	0.365	False
	Kiel	0.099	1	0.105	0.746	0.001	0.127	0.840	False
	Magdeburg	3.254	1	3.453	0.065	0.018	1.189	0.148	False
	Sweden	8.790	1	9.329	0.003	0.047	2.589	0.011	True

7105 rows × 8 columns

Load selected comparison model scores

Show code cell source

Hide code cell source

fname = args.scores_folder / f'diff_analysis_scores_{args.model_key}.pkl'
scores_model = pd.read_pickle(fname)
scores_model

	model	VAE
	var	SS	DF	F	p-unc	np2	-Log10 pvalue	qvalue	rejected
protein groups	Source
A0A024QZX5;A0A087X1N8;P35237	AD	1.049	1	7.630	0.006	0.038	2.201	0.018	True
	age	0.007	1	0.052	0.821	0.000	0.086	0.881	False
	Kiel	0.271	1	1.972	0.162	0.010	0.791	0.266	False
	Magdeburg	0.462	1	3.362	0.068	0.017	1.166	0.133	False
	Sweden	1.666	1	12.119	0.001	0.060	3.209	0.002	True
...	...	...	...	...	...	...	...	...	...
S4R3U6	AD	1.869	1	3.758	0.054	0.019	1.267	0.109	False
	age	0.578	1	1.163	0.282	0.006	0.549	0.411	False
	Kiel	2.571	1	5.167	0.024	0.026	1.617	0.056	False
	Magdeburg	2.381	1	4.786	0.030	0.024	1.524	0.067	False
	Sweden	17.034	1	34.241	0.000	0.152	7.681	0.000	True

7105 rows × 8 columns

Combined scores

show only selected statistics for comparsion

Show code cell source

Hide code cell source

scores = scores_model.join(scores_baseline, how='outer')[[args.baseline, args.model_key]]
scores = scores.loc[:, pd.IndexSlice[scores.columns.levels[0].to_list(),
                                     args.selected_statistics]]
scores

	model	PI				VAE
	var	p-unc	-Log10 pvalue	qvalue	rejected	p-unc	-Log10 pvalue	qvalue	rejected
protein groups	Source
A0A024QZX5;A0A087X1N8;P35237	AD	0.590	0.229	0.728	False	0.006	2.201	0.018	True
	Kiel	0.060	1.224	0.140	False	0.162	0.791	0.266	False
	Magdeburg	0.002	2.657	0.010	True	0.068	1.166	0.133	False
	Sweden	0.000	4.169	0.000	True	0.001	3.209	0.002	True
	age	0.713	0.147	0.817	False	0.821	0.086	0.881	False
...	...	...	...	...	...	...	...	...	...
S4R3U6	AD	0.533	0.273	0.680	False	0.054	1.267	0.109	False
	Kiel	0.746	0.127	0.840	False	0.024	1.617	0.056	False
	Magdeburg	0.065	1.189	0.148	False	0.030	1.524	0.067	False
	Sweden	0.003	2.589	0.011	True	0.000	7.681	0.000	True
	age	0.216	0.666	0.365	False	0.282	0.549	0.411	False

7105 rows × 8 columns

Models in comparison (name mapping)

Show code cell source

Hide code cell source

models = pimmslearn.nb.Config.from_dict(
    pimmslearn.pandas.index_to_dict(scores.columns.get_level_values(0)))
vars(models)

{'PI': 'PI', 'VAE': 'VAE'}

Describe scores

Show code cell source

Hide code cell source

scores.describe()

model	PI			VAE
var	p-unc	-Log10 pvalue	qvalue	p-unc	-Log10 pvalue	qvalue
count	7,105.000	7,105.000	7,105.000	7,105.000	7,105.000	7,105.000
mean	0.261	2.486	0.338	0.224	3.356	0.277
std	0.304	5.378	0.332	0.294	6.304	0.320
min	0.000	0.000	0.000	0.000	0.000	0.000
25%	0.004	0.327	0.015	0.000	0.409	0.002
50%	0.119	0.925	0.238	0.058	1.235	0.116
75%	0.471	2.440	0.628	0.390	3.355	0.520
max	1.000	143.804	1.000	1.000	86.570	1.000

One to one comparison of by feature:

Show code cell source

Hide code cell source

scores = scores.loc[pd.IndexSlice[:, args.target], :]
scores.to_excel(writer, 'scores', **writer_args)
scores

/tmp/ipykernel_102860/3761369923.py:2: FutureWarning: Starting with pandas version 3.0 all arguments of to_excel except for the argument 'excel_writer' will be keyword-only.
  scores.to_excel(writer, 'scores', **writer_args)

	model	PI				VAE
	var	p-unc	-Log10 pvalue	qvalue	rejected	p-unc	-Log10 pvalue	qvalue	rejected
protein groups	Source
A0A024QZX5;A0A087X1N8;P35237	AD	0.590	0.229	0.728	False	0.006	2.201	0.018	True
A0A024R0T9;K7ER74;P02655	AD	0.059	1.228	0.139	False	0.031	1.503	0.070	False
A0A024R3W6;A0A024R412;O60462;O60462-2;O60462-3;O60462-4;O60462-5;Q7LBX6;X5D2Q8	AD	0.114	0.944	0.230	False	0.370	0.432	0.501	False
A0A024R644;A0A0A0MRU5;A0A1B0GWI2;O75503	AD	0.524	0.281	0.672	False	0.258	0.589	0.382	False
A0A075B6H7	AD	0.109	0.964	0.223	False	0.007	2.165	0.019	True
...	...	...	...	...	...	...	...	...	...
Q9Y6R7	AD	0.175	0.756	0.315	False	0.175	0.756	0.284	False
Q9Y6X5	AD	0.050	1.305	0.121	False	0.211	0.675	0.328	False
Q9Y6Y8;Q9Y6Y8-2	AD	0.083	1.079	0.181	False	0.083	1.079	0.156	False
Q9Y6Y9	AD	0.302	0.520	0.464	False	0.503	0.299	0.624	False
S4R3U6	AD	0.533	0.273	0.680	False	0.054	1.267	0.109	False

1421 rows × 8 columns

And the descriptive statistics of the numeric values:

Show code cell source

Hide code cell source

scores.describe()

model	PI			VAE
var	p-unc	-Log10 pvalue	qvalue	p-unc	-Log10 pvalue	qvalue
count	1,421.000	1,421.000	1,421.000	1,421.000	1,421.000	1,421.000
mean	0.254	1.405	0.336	0.236	1.597	0.296
std	0.292	1.623	0.319	0.289	1.828	0.313
min	0.000	0.001	0.000	0.000	0.000	0.000
25%	0.011	0.352	0.036	0.007	0.390	0.019
50%	0.125	0.904	0.247	0.084	1.074	0.157
75%	0.444	1.960	0.605	0.407	2.165	0.536
max	0.997	23.616	0.998	1.000	20.471	1.000

and the boolean decision values

Show code cell source

Hide code cell source

scores.describe(include=['bool', 'O'])

model	PI	VAE
var	rejected	rejected
count	1421	1421
unique	2	2
top	False	False
freq	1027	935

Load frequencies of observed features

Show code cell source

Hide code cell source

freq_feat = pd.read_csv(args.freq_features_observed, index_col=0)
freq_feat.columns = pd.MultiIndex.from_tuples([('data', 'frequency'),])
freq_feat

	data
	frequency
protein groups
A0A024QZX5;A0A087X1N8;P35237	186
A0A024R0T9;K7ER74;P02655	195
A0A024R3W6;A0A024R412;O60462;O60462-2;O60462-3;O60462-4;O60462-5;Q7LBX6;X5D2Q8	174
A0A024R644;A0A0A0MRU5;A0A1B0GWI2;O75503	196
A0A075B6H7	91
...	...
Q9Y6R7	197
Q9Y6X5	173
Q9Y6Y8;Q9Y6Y8-2	197
Q9Y6Y9	119
S4R3U6	126

1421 rows × 1 columns

Compare shared features

Show code cell source

Hide code cell source

scores_common = (scores
                 .dropna()
                 .reset_index(-1, drop=True)
                 ).join(
    freq_feat, how='left'
)
scores_common

	PI				VAE				data
	p-unc	-Log10 pvalue	qvalue	rejected	p-unc	-Log10 pvalue	qvalue	rejected	frequency
protein groups
A0A024QZX5;A0A087X1N8;P35237	0.590	0.229	0.728	False	0.006	2.201	0.018	True	186
A0A024R0T9;K7ER74;P02655	0.059	1.228	0.139	False	0.031	1.503	0.070	False	195
A0A024R3W6;A0A024R412;O60462;O60462-2;O60462-3;O60462-4;O60462-5;Q7LBX6;X5D2Q8	0.114	0.944	0.230	False	0.370	0.432	0.501	False	174
A0A024R644;A0A0A0MRU5;A0A1B0GWI2;O75503	0.524	0.281	0.672	False	0.258	0.589	0.382	False	196
A0A075B6H7	0.109	0.964	0.223	False	0.007	2.165	0.019	True	91
...	...	...	...	...	...	...	...	...	...
Q9Y6R7	0.175	0.756	0.315	False	0.175	0.756	0.284	False	197
Q9Y6X5	0.050	1.305	0.121	False	0.211	0.675	0.328	False	173
Q9Y6Y8;Q9Y6Y8-2	0.083	1.079	0.181	False	0.083	1.079	0.156	False	197
Q9Y6Y9	0.302	0.520	0.464	False	0.503	0.299	0.624	False	119
S4R3U6	0.533	0.273	0.680	False	0.054	1.267	0.109	False	126

1421 rows × 9 columns

Annotate decisions in Confusion Table style:

Show code cell source

Hide code cell source

def annotate_decision(scores, model, model_column):
    return scores[(model_column, 'rejected')].replace({False: f'{model} (no) ', True: f'{model} (yes)'})


annotations = None
for model, model_column in models.items():
    if annotations is not None:
        annotations += ' - '
        annotations += annotate_decision(scores_common,
                                         model=model, model_column=model_column)
    else:
        annotations = annotate_decision(
            scores_common, model=model, model_column=model_column)
annotations.name = 'Differential Analysis Comparison'
annotations.value_counts()

Differential Analysis Comparison
PI (no)  - VAE (no)    878
PI (yes) - VAE (yes)   337
PI (no)  - VAE (yes)   149
PI (yes) - VAE (no)     57
Name: count, dtype: int64

List different decisions between models

Show code cell source

Hide code cell source

mask_different = (
    (scores_common.loc[:, pd.IndexSlice[:, 'rejected']].any(axis=1))
    & ~(scores_common.loc[:, pd.IndexSlice[:, 'rejected']].all(axis=1))
)
_to_write = scores_common.loc[mask_different]
_to_write.to_excel(writer, 'differences', **writer_args)
logger.info("Writen to Excel file under sheet 'differences'.")
_to_write

/tmp/ipykernel_102860/1417621106.py:6: FutureWarning: Starting with pandas version 3.0 all arguments of to_excel except for the argument 'excel_writer' will be keyword-only.
  _to_write.to_excel(writer, 'differences', **writer_args)
root - INFO     Writen to Excel file under sheet 'differences'.

	PI				VAE				data
	p-unc	-Log10 pvalue	qvalue	rejected	p-unc	-Log10 pvalue	qvalue	rejected	frequency
protein groups
A0A024QZX5;A0A087X1N8;P35237	0.590	0.229	0.728	False	0.006	2.201	0.018	True	186
A0A075B6H7	0.109	0.964	0.223	False	0.007	2.165	0.019	True	91
A0A075B6H9	0.385	0.414	0.550	False	0.020	1.710	0.047	True	189
A0A075B6J9	0.052	1.285	0.125	False	0.009	2.063	0.024	True	156
A0A075B6Q5	0.510	0.292	0.662	False	0.003	2.459	0.011	True	104
...	...	...	...	...	...	...	...	...	...
Q9UKB5	0.010	1.984	0.034	True	0.103	0.988	0.186	False	148
Q9UNW1	0.010	1.999	0.033	True	0.965	0.016	0.978	False	171
Q9UP79	0.330	0.481	0.494	False	0.000	4.546	0.000	True	135
Q9UQ52	0.058	1.233	0.138	False	0.000	3.377	0.002	True	188
Q9Y6C2	0.751	0.125	0.843	False	0.013	1.874	0.034	True	119

206 rows × 9 columns

Plot qvalues of both models with annotated decisions

Prepare data for plotting (qvalues)

Show code cell source

Hide code cell source

var = 'qvalue'
to_plot = [scores_common[v][var] for v in models.values()]
for s, k in zip(to_plot, models.keys()):
    s.name = k.replace('_', ' ')
to_plot.append(scores_common['data'])
to_plot.append(annotations)
to_plot = pd.concat(to_plot, axis=1)
to_plot

	PI	VAE	frequency	Differential Analysis Comparison
protein groups
A0A024QZX5;A0A087X1N8;P35237	0.728	0.018	186	PI (no) - VAE (yes)
A0A024R0T9;K7ER74;P02655	0.139	0.070	195	PI (no) - VAE (no)
A0A024R3W6;A0A024R412;O60462;O60462-2;O60462-3;O60462-4;O60462-5;Q7LBX6;X5D2Q8	0.230	0.501	174	PI (no) - VAE (no)
A0A024R644;A0A0A0MRU5;A0A1B0GWI2;O75503	0.672	0.382	196	PI (no) - VAE (no)
A0A075B6H7	0.223	0.019	91	PI (no) - VAE (yes)
...	...	...	...	...
Q9Y6R7	0.315	0.284	197	PI (no) - VAE (no)
Q9Y6X5	0.121	0.328	173	PI (no) - VAE (no)
Q9Y6Y8;Q9Y6Y8-2	0.181	0.156	197	PI (no) - VAE (no)
Q9Y6Y9	0.464	0.624	119	PI (no) - VAE (no)
S4R3U6	0.680	0.109	126	PI (no) - VAE (no)

1421 rows × 4 columns

List of features with the highest difference in qvalues

Show code cell source

Hide code cell source

# should it be possible to run not only RSN?
to_plot['diff_qvalue'] = (to_plot[str(args.baseline)] - to_plot[str(args.model_key)]).abs()
to_plot.loc[mask_different].sort_values('diff_qvalue', ascending=False)

	PI	VAE	frequency	Differential Analysis Comparison	diff_qvalue
protein groups
P52758	0.000	0.999	119	PI (yes) - VAE (no)	0.999
A0A087WU43;A0A087WX17;A0A087WXI5;P12830;P12830-2	0.989	0.000	134	PI (no) - VAE (yes)	0.989
O15204;O15204-2	0.982	0.013	156	PI (no) - VAE (yes)	0.969
P22692;P22692-2	0.998	0.041	170	PI (no) - VAE (yes)	0.957
Q504Y2	0.970	0.023	96	PI (no) - VAE (yes)	0.947
...	...	...	...	...	...
P00740;P00740-2	0.053	0.043	197	PI (no) - VAE (yes)	0.010
K7ERG9;P00746	0.052	0.042	197	PI (no) - VAE (yes)	0.010
B3KTY4;Q9H156;Q9H156-2	0.045	0.053	141	PI (yes) - VAE (no)	0.008
P26572	0.056	0.049	194	PI (no) - VAE (yes)	0.006
Q16706	0.052	0.048	195	PI (no) - VAE (yes)	0.004

206 rows × 5 columns

Differences plotted with created annotations

Show code cell source

Hide code cell source

figsize = (4, 4)
size = 5
fig, ax = plt.subplots(figsize=figsize)
x_col = to_plot.columns[0]
y_col = to_plot.columns[1]
ax = sns.scatterplot(data=to_plot,
                     x=x_col,
                     y=y_col,
                     s=size,
                     hue='Differential Analysis Comparison',
                     ax=ax)
_ = ax.legend(fontsize=fontsize,
              title_fontsize=fontsize,
              markerscale=0.4,
              title='',
              )
ax.set_xlabel(f"qvalue for {x_col}")
ax.set_ylabel(f"qvalue for {y_col}")
ax.hlines(0.05, 0, 1, color='grey', linestyles='dotted')
ax.vlines(0.05, 0, 1, color='grey', linestyles='dotted')
sns.move_legend(ax, "upper right")
files_out[f'diff_analysis_comparision_1_{args.model_key}'] = (
    args.out_folder /
    f'diff_analysis_comparision_1_{args.model_key}')
fname = files_out[f'diff_analysis_comparision_1_{args.model_key}']
pimmslearn.savefig(fig, name=fname)

pimmslearn.plotting - INFO     Saved Figures to runs/alzheimer_study/diff_analysis/AD/PI_vs_VAE/diff_analysis_comparision_1_VAE

• also showing how many features were measured (“observed”) by size of circle

Show code cell source

Hide code cell source

fig, ax = plt.subplots(figsize=figsize)
ax = sns.scatterplot(data=to_plot,
                     x=to_plot.columns[0],
                     y=to_plot.columns[1],
                     size='frequency',
                     s=size,
                     sizes=(5, 20),
                     hue='Differential Analysis Comparison')
_ = ax.legend(fontsize=fontsize,
              title_fontsize=fontsize,
              markerscale=0.6,
              title='',
              )
ax.set_xlabel(f"qvalue for {x_col}")
ax.set_ylabel(f"qvalue for {y_col}")
ax.hlines(0.05, 0, 1, color='grey', linestyles='dotted')
ax.vlines(0.05, 0, 1, color='grey', linestyles='dotted')
sns.move_legend(ax, "upper right")
files_out[f'diff_analysis_comparision_2_{args.model_key}'] = (
    args.out_folder / f'diff_analysis_comparision_2_{args.model_key}')
pimmslearn.savefig(
    fig, name=files_out[f'diff_analysis_comparision_2_{args.model_key}'])

pimmslearn.plotting - INFO     Saved Figures to runs/alzheimer_study/diff_analysis/AD/PI_vs_VAE/diff_analysis_comparision_2_VAE

Only features contained in model

• this block exist due to a specific part in the ALD analysis of the paper

Show code cell source

Hide code cell source

scores_model_only = scores.reset_index(level=-1, drop=True)
_diff = scores_model_only.index.difference(scores_common.index)
if not _diff.empty:
    scores_model_only = (scores_model_only
                         .loc[
                             _diff,
                             args.model_key]
                         .sort_values(by='qvalue', ascending=True)
                         .join(freq_feat.squeeze().rename(freq_feat.columns.droplevel()[0])
                               )
                         )
    display(scores_model_only)
else:
    scores_model_only = None
    logger.info("No features only in new comparision model.")

if not _diff.empty:
    scores_model_only.to_excel(writer, 'only_model', **writer_args)
    display(scores_model_only.rejected.value_counts())
    scores_model_only_rejected = scores_model_only.loc[scores_model_only.rejected]
    scores_model_only_rejected.to_excel(
        writer, 'only_model_rejected', **writer_args)

root - INFO     No features only in new comparision model.

DISEASES DB lookup

Query diseases database for gene associations with specified disease ontology id.

Show code cell source

Hide code cell source

data = pimmslearn.databases.diseases.get_disease_association(
    doid=args.disease_ontology, limit=10000)
data = pd.DataFrame.from_dict(data, orient='index').rename_axis('ENSP', axis=0)
data = data.rename(columns={'name': args.annotaitons_gene_col}).reset_index(
).set_index(args.annotaitons_gene_col)
data

pimmslearn.databases.diseases - WARNING  There are more associations available

	ENSP	score
None
APOE	ENSP00000252486	5.000
PSEN1	ENSP00000326366	5.000
PSEN2	ENSP00000355747	5.000
APP	ENSP00000284981	5.000
TREM2	ENSP00000362205	4.825
...	...	...
CARMIL1	ENSP00000331983	0.681
CENPJ	ENSP00000371308	0.681
ERP27	ENSP00000266397	0.681
ZNF585B	ENSP00000433773	0.681
KIR3DL2	ENSP00000325525	0.681

10000 rows × 2 columns

Shared features

ToDo: new script -> DISEASES DB lookup

Show code cell source

Hide code cell source

feat_name = scores.index.names[0]  # first index level is feature name
if args.annotaitons_gene_col in scores.index.names:
    logger.info(f"Found gene annotation in scores index:  {scores.index.names}")
else:
    logger.info(f"No gene annotation in scores index:  {scores.index.names}"
                " Exiting.")
    import sys
    sys.exit(0)

root - INFO     No gene annotation in scores index:  ['protein groups', 'Source'] Exiting.
/home/runner/work/pimms/pimms/project/.snakemake/conda/924ec7e362d761ecf0807b9074d79999_/lib/python3.12/site-packages/IPython/core/interactiveshell.py:3707: UserWarning: To exit: use 'exit', 'quit', or Ctrl-D.
  warn("To exit: use 'exit', 'quit', or Ctrl-D.", stacklevel=1)

An exception has occurred, use %tb to see the full traceback.

SystemExit: 0

Show code cell source

Hide code cell source

gene_to_PG = (scores.droplevel(
    list(set(scores.index.names) - {feat_name, args.annotaitons_gene_col})
)
    .index
    .to_frame()
    .reset_index(drop=True)
    .set_index(args.annotaitons_gene_col)
)
gene_to_PG.head()

Show code cell source

Hide code cell source

disease_associations_all = data.join(
    gene_to_PG).dropna().reset_index().set_index(feat_name).join(annotations)
disease_associations_all

only by model

Show code cell source

Hide code cell source

idx = disease_associations_all.index.intersection(scores_model_only.index)
disease_assocications_new = disease_associations_all.loc[idx].sort_values(
    'score', ascending=False)
disease_assocications_new.head(20)

Show code cell source

Hide code cell source

mask = disease_assocications_new.loc[idx, 'score'] >= 2.0
disease_assocications_new.loc[idx].loc[mask]

Only by model which were significant

Show code cell source

Hide code cell source

idx = disease_associations_all.index.intersection(
    scores_model_only_rejected.index)
disease_assocications_new_rejected = disease_associations_all.loc[idx].sort_values(
    'score', ascending=False)
disease_assocications_new_rejected.head(20)

Show code cell source

Hide code cell source

mask = disease_assocications_new_rejected.loc[idx, 'score'] >= 2.0
disease_assocications_new_rejected.loc[idx].loc[mask]

Shared which are only significant for by model

mask = (scores_common[(str(args.model_key), 'rejected')] & mask_different)
mask.sum()

Show code cell source

Hide code cell source

idx = disease_associations_all.index.intersection(mask.index[mask])
disease_assocications_shared_rejected_by_model = (disease_associations_all.loc[idx].sort_values(
    'score', ascending=False))
disease_assocications_shared_rejected_by_model.head(20)

Show code cell source

Hide code cell source

mask = disease_assocications_shared_rejected_by_model.loc[idx, 'score'] >= 2.0
disease_assocications_shared_rejected_by_model.loc[idx].loc[mask]

Only significant by RSN

mask = (scores_common[(str(args.baseline), 'rejected')] & mask_different)
mask.sum()

Show code cell source

Hide code cell source

idx = disease_associations_all.index.intersection(mask.index[mask])
disease_assocications_shared_rejected_by_RSN = (
    disease_associations_all
    .loc[idx]
    .sort_values('score', ascending=False))
disease_assocications_shared_rejected_by_RSN.head(20)

Show code cell source

Hide code cell source

mask = disease_assocications_shared_rejected_by_RSN.loc[idx, 'score'] >= 2.0
disease_assocications_shared_rejected_by_RSN.loc[idx].loc[mask]

Write to excel

Show code cell source

Hide code cell source

disease_associations_all.to_excel(
    writer, sheet_name='disease_assoc_all', **writer_args)
disease_assocications_new.to_excel(
    writer, sheet_name='disease_assoc_new', **writer_args)
disease_assocications_new_rejected.to_excel(
    writer, sheet_name='disease_assoc_new_rejected', **writer_args)

Outputs

Show code cell source

Hide code cell source

writer.close()
files_out