TKosfeld/Genome-Viz

3300 Term Project for Processing and Visualization of Genomic Data

Genome-Viz

3300 Term Project for Processing and Visualization of Genomic Data

MergePackage

TSV file data handler operating using argparse command line.

Example

python3 Genome_Viz.py test_data/ test -d MPXVInfected2wks MPXVInfected8wks -b 0 1  

Sprint 1 Burndown

Tasks:

• Import and utilize Argparse Module to allow flexibility in file selection.
• Evaluate Program efficiency using tm.time.
• Implement file identification, selection, and sorting function under title: loadData().
• Based on file selection, create and identifier row denote sample status as infected or naive.
• Implement a labeling function, getNames(), to appropriately track and identify files in the final data frame.
• Initialize an empty dictionary to hold concatenated data.
• Trim the dictionary filling function fillDict().
• FillDict() should ignore empty sequence entries and disregard non-data columns while processing data for entry.
• FillDict() should be able to process flawed files with unexpected formating.
• FillDict() should store each sequence in a key relevant to the specific file the sequence was read from.
• FillDict() should return a dictionary of dictionaries upon completion.
• Sprint 1 package should print a running time and completely dictionary of dictionaries.

Sprint 2 Burndown

Tasks:

• For current coherence, single cohort data frames should be recorded and expressed as .csv files.
• For coherence the prototype should notify the user when single cohort files have been processed into a singular dictionary.
• Argparse functionality should be added to determine location and naming of output single cohort data file.
• The .csv should be written to by line, with column headers of filenames serving as the first written row.
• Column specific exposure based labels should be assigned to each header.
• In an effort to prevent overpopulation of irrelevant data, the .csv writer should require a certain amount of samples expressed for a TCR sequence to be recognized.
• As in projected dataframe expression, each row should be declared as an empty list and filled from relevant dictionary entries.
• Relevant columns lacking data dictionary entries for a certain sequence should be represented with 0’s in the row.
• Relevant efficiency and timekeeping functions should be updated to accommodate new workloads.

Sprint 3 Burndown

Tasks: (6 Tasks, 2 Weeks)

• Line separation should be mutable between \t and ,.
• Column specification should be mutable through argparse specification.
• Filetype should be mutable through argparse specification.
• Institute filtering of relevant data for classification.
• Mutable line separation variable is employed on a line by line basis when processing data.
• Dataset specification will allow for multiple datasets to be joined from a single directory specified in the command line.

Sprint 4 Burndown

Tasks: (9 Tasks, 2 Weeks)

• Cohort data frames should be processed and interpreted as floats.
• Cohort representations should be created through concatenation of all labeled files.
• The joining of this genomic data should be done by averaging sequence expression over the cohort.
• For each shared sequence between the two cohorts I want to calculate a percentage deviation.
• To avoid skewed data the percentage deviation should be normalized between 0 and 100%/
• The mean of these normalized percentages will be displayed as the overall deviation between cohorts.
• These normalized percentages should be appended to the dataframe.
• The sequences with the largest deviations should be presented as high-impact characteristic sequences.
• The program should display the 50 highest impact sequences and their mean representation in each cohort.

Sprint 5 Burndown

Tasks: (12 Tasks, 1 Week)

• The program needs to import dash and dash_bio for their visualization capabilities.
• The program should have an html template and dash app for server creation.
• The dash app object should be created with coherent text and color.
• The visualization process should receive the data of the 50 highest impact sequences, sorted by cohort deviation.
• The cohorts should be visualized side by side in a clustergram to properly visualize differences.
• The clustergram colors should be distinct and conductive towards differentiation.
• The clustergram should identify possible clusters and recurrent behavior.
• Cohort columns should be labelled appropriately.
• Rows should be identified by sequence.
• The graph should be fitted to the screen while being large enough to distinguish individual rows.
• The graph should be added to the template in the app object.
• The server should be run without debugging to maintain an interactive single-run visualization.