fvalle1 / master_thesis

Network analysis on TCGA data

Pipeline

Pipeline contains all the necessary steps to run a complete analysis

Run statistics

import tacos
tacos.statistics()
tacos.tacos.shuffleLabels()

Make bipartite network

from sbmtm import sbmtm
model = sbmtm()
model.make_graph_from_BoW(df)
model.save_graph("graph.xml.gz")

Run stochastic block model

interactively

docker run -it -p8888:8888 -v $(PWD):/home/filippo/files --entrypoint /bin/bash fvalle01/hsbm 2
jupyter notebook --ip0.0.0.0

as executable

docker run -it -v $(PWD):/home/filippo/files/ fvalle01/hsbm:standard 2

Note there must be a graph.xml.gz file in current working directory

TCGA

TCGA get manifest

TCGA_GetManifest creates a file called manifest.txt that is useful to download data using gdc-client download -m manifest.txt command

TCGA API

TCGA_API is useful to retrieve informations about a single file (sample)

Table

Table creation

Table_Creation reads a folder with the data downloaded with gdc-client and creates a mainTable.csv dataset

Table mining

Table_Mining is useful to extract means and vars per tissues. It stores a meanVariances.csv file useful in next step

Table Analyzer

Table_Analyzer is useful to plot expression per tissues histograms and FPKM means and distributions. Here it mean versus occurrence is created.

Table by gene type

Table_ByGeneType is useful to do analysis separating protein coding from non coding excetera

Table by tissue

Table_ByTissue perform analysis looking at different tissues

Table protein coding

Table_ProteinCoding isolate a mainTable.csv file with only protein coding genes. Creates a genes_pc.txt file with all ENSG-id that are related to a protein-coding gene

Table microRNA

Table_microRNA as above but with microRNA

Table Analyzer

Table_tfidf perform tf-idf analysis

hSBM

topics

hSBM_topics analyse topic composition

clusters

hSBM_clusters bench the performance in clustering samples

topic_dist

hSBM_topic-dist analyse topic distribution inside samples

Analysis

Use Tool for Analyse COmponents Systems

Zipf

Use Zipf.ipynb to plot Zipf and U (occurrence distribution)

Heaps

Use Heaps.ipynb to plot Heaps distibution