The GitHub repository http:/github.com/ruppinlab/ProcessTrialtrove contains two programs usful for processing data from the clinical trial repository Trialtrove. The code will be directly useful only for those who have a Trialtrove license. For others, the code should be interpreted as pseudocode that clarifies certain methods we are using to analyze Trialtrove data.

To analyze Trialtrove data in a frozen manner, one must download a set of trials with all fields as an Excel file. In general, the trials may be in multiple disjoint Excel files. For some purposes, it useful to convert the Excel format to ASCII text, which makes it possible to use search tools such as awk and grep on UNIX. However, the Excel files may contain many non-ASCII characters and an assotment of whitespace characters within each cell.

Therefore, trialtrove_processing_wcomments.py and character_conversion_table.py are used to do the the conversion from Excel to ASCII Usage: python -c 'import trialtrove_processing_wcomments; trialtrove_processing_wcomments.process("input.xlsx","output.txt") where the file names input.xlsx and output.txt may be replaced.

For the project entitled "Outcome Differences by Sex in Oncology Clinical Trials", Ashwin wrote the program query_sex_differences.py which uses the auxiliary file sex_trial_keyword_list.py. The purpose of query_sex_differences.py is to find a subset of clinical trials in which the Trialtrove curators may have found a comparison of males vs. females for outcomes or side effects. Usage: python3 query_sex_differences.py

The inputs are specified in the program with lines such as: file_1 = pd.read_csv("/data/Vegesna_Schaffer/TrialTrove/freeze2/TrialTrove_Oncology_File1_20221223.txt",sep='\t', lineterminator='\n') and combined with the line data = pd.concat([file_1, file_2,file_3, file_4, file_5, file_6, file_7, file_8, file_9, file_10,file_11, file_12, file_13, file_14, file_15, file_16, file_17,file_18, file_19, file_20, file_21, file_22, file_23, file_24,file_25, file_26, file_27, file_28, file_29, file_30, file_31,file_32]) and anyone else using this program would need to change those lines to specify the full paths of the input files and the correct number of files to combine.

The primary output of query_sex_differences.py is an Excel file with the following columns: Trial ID: Unique positiver integer assigned by Trialtrove to each trial Gender Term: The token that may refer to males or females Comparison Term: The token or substring (such as "vs.) that suggests a comparison has been done Context: The context around the sex term that may include a comparison Column: Either Trial Notes or Trial Results (because these are the two Trailtrove columns that contain results and where comparisons by sex may be found)

Both trialtrove_processing_wcomments.py and python3 query_sex_differences.py require the pandas package of python

Contributors: Ashwin Kammula and Alejandro Schaffer Contact: Alejandro Schaffer (alejandro.schaffer@nih.gov)