Wiki¶

Data analysis background¶

The data was originally analysed by overseas collaborators (Hans mollenkopf@mpiib-berlin.mpg.de and Natalie natalie.nieuwenhuizen@uni-wuerzburg.de). Hans provided .RData objects and .RMD file summarizing his results, which were quite comprehensive (including exploratory analysis, differential expression using limma and gene module enrichment using the R package tmod). However there are very strong batch effects in this dataset that correspond to RNA extraction date in the lab. I therefore redid the analysis with combat batch correction (which was only possible by correcting the astrocyte and neuron samples separately). They were therefore analysed sepearately downstream.

Data preprocessing¶

Hans performed background correction (normexp method) and normalization (quantile). The data was supplied to me as .RData objects (raw and normalized). I added ComBat batch correction.

Downstream analysis (see attached powerpoint)¶

PCA, tSNE multivariate

Differential expression analysis between groups:
Astrocytes: not infected vs. H37Rv; not infected vs BC; BCG vs H37Rv
Neurons: not infected vs. H37Rv

Transcriptional module testing: using R package tmod

Signaling Pathway Impact Analysis (SPIA) for neurons (NI vs. RV) and astrocytes (NI vs. RV).