We expand MAPS strategy to 1,080 FDA approved drugs for large scale target deconvolution for drug repurposing and discovery

• Figures: Containing all main figures and supplementary figures with every sub-figures
• manuscript: Including every versions of manuscripts modified by C.S.H.T and Bin Liao
• script: The main part of this project which comprise core codes, input/output files, being utilized files and etc.
• Source data: Experimental validation data
• Table: Preliminary data information of this project (it has not been updated for long time, but latest table can be integrated from script file)

• All the Figures and supplementary figures have been listed here and the script file here involves codes used to plot some sub-figures of every figures
• Since there are a series of update for figures, maybe part of sub-figures do not match the figures in .ai files even though they are in same files
• The Figure Candidate file contain the figures that are not be performed in main/supplementary figures (part of them have actually been displayed in figures)

• MS_files: Oringnal MS files derived from 480 and Lumos
• revised result: Data processing and normalization result of MS data
• combine_group_matrix: Pooling drug matrix of each group
• resource data: All public data employed to integrate, map and calculate in house data
• results: Basic calculation result (FC, pvalue and rank based on pvalue) for every drugs in each group
• notebooks: Core codes for data processing, computation and visualization
• pkl_files: Transition files generated during data calculations via Python
• output_file: Generated key information table from data integration and calculation
• auto search articles: PubMed search result for significant drug-protein pairs (fc > 0.1 & pvalue < 0.05)
• drugs: Structure images of 44 drugs that identified more than 10 significant proteins
• The .py files include all function codes called by notebooks code

• All the codes are programmed by Python (version 3.11)
• Following main packages will be implemented:

adjust_text 0.8, Bio 1.6.2, biomart 0.9.2, imbalanced-learn 0.11.0
matplotlib 3.7.2, multiprocessing 0.70.14, networkx 3.1, numpy 1.24.3
pandas 1.5.3, pickle 0.7.5, seaborn 0.11.2, scipy 1.11.1, scikit-learn 1.3.0

• data process&QC visilization: Basic data cleaning (i.e. filtering low confidence proteins based on PSM and median normalization) for generation of files in revised result
• data normalization: Calculation, combination and integration of in total 45 groups from different two MS instrucments. During the pvalue calculation, we employed DBSCAN to overcome the problem that two drugs target to same proteins within same groups. The generated result is located in results file and we got an output table 1. sig_all_drug_prot.xlsx which include all drug-protein pairs information across 45 groups with the threhold fc > 0.1 & pvalue < 0.05
• article auto search for drug-protein pairs: With the utilization of Bio package and PubMed, we achieve article auto search for all drug-protein pairs in 1. sig_all_drug_prot.xlsx. Then, we also combine the information from DrugBank and STITCH to maximize the source of known drug-protein pairs and the result can be checked in 2. drug_target_report.xlsx
• threshold analysis: Here, we set out different pairwised fc and pvalue but can not obtain an idea threshold combination. Therefore, the other approches need to be considered (such as machine learning) to discover more confident drug-protein pairs
• balancedRF: We employed a machine learning algorithm namely Balanced Radom Forest Classifier (BRFC) to predict high-confidence drug-protein pairs, comprehensively including various features. The features and prediction result can be checked in 3. output_table_features.csv
• network_assembly: Network analysis based on the result of 3. output_table_features.csv
• volcano plot of summary data: Drug-centric and protein-centric volcano plot to visualize significant proteins and drugs respectively
• public resource: In total 1,037 significant proteins were clasified based on database, like Pharos, Panther, HPA, to descript the characteristics of these proteins. The combination result generate table 4. public resource data.xlsx