Exploratory Data Analysis and Visualization System for Biostatistical Research

data/

Please note: For the public repository, data/ has been omitted to respect data privacy/licensing.

The data/ directory contains the datasets used in the project. It includes two subdirectories:

original/ Original data files imported from outside this repository

• adni/
• aric/
• calcium/
• other/

processed/ Interim data files generated manually or by a script within this repository

• adni/
• aric/
• other/

assets/

The assets/ directory contains final, presentation-ready tables, figures, and slides

tables/ Demographic characteristics tables, summary statistics tables, etc.

• adni/
• aric/

figures/ Saved figures in PNG (raster/pixel) and PDF (vector) formats by interactive Dash apps

• adni/
• aric/

slides/ Summary slides

src/

The src/ directory contains the source code for this project. It is organized into the following subdirectories:

src/lib/

This directory contains the library code for the project. The utility functions are organized into the following categories:

general.py General utility functions
• get_stage_list() returns the list of names of stages stratified by biomarkers Ab42, amyloid PET, (p-Tau or t-Tau)
stats.py Statistical analysis
• demographics_characteristics() computes a summary statistics table of the study population
• multiple_linear_regression() fits a multiple linear regression model and returns model statistics
• cluster_corr_df() hierarchially clusters a correlation matrix
  • get_linkage_methods() returns list of available linkage methods for hierarchical clustering
  • get_cluster_criteria() returns list of available cluster criteria for hierarchical clustering
• remove_diagonal() masks the diagonal of a square matrix with NaN
• fill_mirror() fills a triangular matrix to a square matrix with its transpose
• mask_outlier() returns a mask that removes outliers when applied. Outliers are determined by Local Outlier Factor
• corr_remove_outliers() computes the outlier-removed correlation coefficient for each pair of variables; returns a correlation matrix
dialog.py Tkinter dialogs
• dialog_select_directory() prompts the user to select a directory in a dialog selection window and returns its absolute path
• dialog_select_file() prompts the user to select a file in a dialog selection window and returns its absolute path
plotly.py Modifications to Plotly figure objects
• standard_layout() configures a standard layout for plot template, axes, and fonts
• add_box() draws a box plot on top of a strip plot
• add_pairwise_comparison() annotates pairwise comparison results on top of a strip plot or box plot
  • annotation_t_test() computes the p-value from independent-sample t-test
  • annotation_cohens_d() computes Cohen's d effect size
  • annotation_tukey() performs Tukey's multiple comparison post-hoc to obtain the p-value
r_interface.py Interface to R
• tukey() conducts Tukey's multiple comparison post-hoc in R for a single dependent variable
• tukey_multiple_dvs() conducts Tukey's multiple comparison post-hoc in R sequentially for a list of dependent variables; returns a table of the resultant p-values

src/processing/

This directory contains scripts that process original files and/or processed files to processed files for downstream analyses.

adni/

bmi.ipynb Body Mass Index (BMI)

• Input: VITALS_14Jul2023.csv
• Output: bmi.csv

strem2.ipynb CSF soluble triggering receptor expressed on myeloid cells 2 (sTREM2)

• Input: ADNI_HAASS_WASHU_LAB_13Jul2023.csv
• Output: strem2.csv

demographics.ipynb Basic demographics

• Input: ADNIMERGE_14Jul2023.csv, bmi.csv
• Output: demographics.csv

demographics_tau.ipynb Demographics with tau biomarker data

• Input: ADNIMERGE_14Jul2023.csv, bmi.csv
• Output: demographics_tau.csv

demographics_biomarkers.ipynb Demographics with amyloid and tau biomarker data and stage assignment

• Input: ADNIMERGE_14Jul2023.csv, bmi.csv, strem2.csv
• Output: demographics_biomarkers.csv

lipidomics.ipynb Plasma lipidomics, Meikle lab, longitudinal

• Input: ADMCLIPIDOMICSMEIKLELABLONG_13Jul2023.csv, Lipid_Models_Final.xlsx
• Output: lipidomics.csv, lipidomics_total.csv, lipidomics_dictionary.csv

lipoprotein.ipynb Nightingale NMR analysis of lipoproteins and metabolites

• Input: ADNINIGHTINGALELONG_05_24_21_27Jul2023.csv
• Output: lipoprotein.csv, lipoprotein_dict.csv

somascan.ipynb CSF proteomics SOMAscan 7000+ proteins post-QC, Cruchaga lab

• Input: CruchagaLab_CSF_SOMAscan7k_Protein_matrix_postQC_20230620.csv, ADNI_Cruchaga_lab_CSF_SOMAscan7k_analyte_information_20_06_2023.csv
• Output: somascan.csv, somascan_dict.csv

converters.ipynb Longitudinal decline in cognitive status (CN to MCI, MCI to AD, or CN to AD), excluding participants diagnosed with AD at baseline

• Input: ADNIMERGE_14Jul2023.csv
• Output: converters.csv

converters_to_ad.ipynb Longitudinal decline in cognitive status from CN or MCI to AD, excluding participants diagnosed with AD at baseline

• Input: ADNIMERGE_14Jul2023.csv
• Output: converters_to_ad.csv

aric/

pilot.ipynb Demographics and brain MRI data from the ARIC server for participants included in the pilot study

• Input from ARIC server: ARIC_NP/DATA_NP/Visits/Visit 5/derive54_np.sas7bdat, DATA_NP/Visits/Visit 5/derive_ncs51_np.sas7bdat, DATA_NP/Visits/Visit 1/derive13_np.sas7bdat
• Input: all_eleigible_samples_AS2021_25v3.xlsx, lipoproteins_6_29_23.csv, dictionary.csv
• Output: lipoprotein_list.csv, pilot.csv, demographic_characteristics.csv

pilot_eligible.ipynb Demographics and brain MRI data for ARIC participants eligible under the inclusion criteria

• Input from ARIC server: ARIC_NP/DATA_NP/Visits/Visit 5/derive54_np.sas7bdat, DATA_NP/Visits/Visit 5/derive_ncs51_np.sas7bdat, DATA_NP/Visits/Visit 1/derive13_np.sas7bdat, DATA_NP/Visits/MultiVisit/V5_V11 Longitudinal MRI data/v5_v11_mri_derv_np_240221.sas7bdat
• Input: all_eleigible_samples_AS2021_25v3.xlsx, ARIC_Pilot_Updated_06032022.csv, lipoproteins_6_29_23.csv, dictionary.csv
• Output: lipoprotein_list.csv, pilot.csv, demographic_characteristics.csv

other/

davidson.ipynb Sean Davidson HDL Proteome Watch 2023

• Input: HDL Proteome Watch 2023 Final.xlsx
• Output: hdl_proteome_davidson.csv

src/analysis/

This directory contains Jupyter notebook files that perform analyses.

adni/
• lipidomics_tukey.ipynb ANCOVA followed by Tukey post-hoc to determine which plasma lipids or biomarkers differ significantly between stages
• lipidomics_boxplot.ipynb Distribution of plasma lipids or biomarkers across stages
• survival.ipynb Survival analysis (Kaplan-Meier survival curve, Cox's proportional hazard model) comparing risk of conversion to AD between biomarker groups.
• survival_hdl_ratio.ipynb Survival analysis comparing cognitive decline between tertiles of non-small HDL FC-to-CE ratio.
• somascan_pca.ipynb Clustering of CSF proteins by PCA, followed by linear regression with dependent variable pTau
• somascan_boxplot.ipynb Distribution of CSF proteins across cognitive statuses
• strem2_lipidomics_regression.ipynb Linear regression of CSF sTREM2 on plasma lipids.
• strem2_lipoprotein_regression.ipynb Linear regression of CSF sTREM2 on plasma lipoprotein subclasses.
aric/
calcium/
• calcium_all_sites.ipynb Distribution of calcium measurements compared between Vista and Roche, data from all sites combined
other/
• imagej_particle_results_hdl.ipynb HDL1 and HDL2 particle analysis on EM images using results exported from ImageJ

Publications

The analysis in this repository contributed to the following publications:

• Li, D.; Mantyh, W. G.; Men, L.; Jain, I.; Glittenberg, M.; An, B.; Zhang, L.; Li, L.; for the Alzheimer’s Disease Neuroimaging Initiative. sTREM2 in Discordant CSF Aβ42 and P‐tau181. Alz & Dem Diag Ass & Dis Mo 2025, 17 (1), e70072. https://doi.org/10.1002/dad2.70072.

• Li, D.; An, B.; Men, L.; Glittenberg, M.; Lutsey, P. L.; Mielke, M. M.; Yu, F.; Hoogeveen, R. C.; Gottesman, R.; Zhang, L.; Meyer, M.; Sullivan, K.; Zantek, N.; Alonso, A.; Walker, K. A. The Association of High-Density Lipoprotein Cargo Proteins with Brain Volume in Older Adults in the Atherosclerosis Risk in Communities (ARIC). Journal of Alzheimer’s Disease 2025, 103 (3), 724–734. https://doi.org/10.1177/13872877241305806.

Data Sources

• Alzheimer's Disease Neuroimaging Initiative (ADNI)
• The Atherosclerosis Risk in Communities Study (ARIC)