Fumigatus Microarray Data
Obtained the original data in form of TIGR MeV .anl project files from Jason (~/data/projects/a_fungus/2/ma_data)
- 4 or 5 combined expression arrays (2h has 4 arrays). Combined means that the expression values from dye swaps have been combined into single array in TIGR's MIDAS (so originally ten arrays). (note: I am not 100% on this??).
- Filtering: probes that are absent in any array are removed.
- Significance Test: T Test with p cutoff of 0.01. I re-ran this with p cutoff of 0.05 to include what i consider positive controls (i.e. SidA Afu2g07680)
- Small range in the fold expression (0.6 - -0.6). This does not seem normal. This also creates a problem for the T test. Standard T tests can identify genes with small, consistent fold change, but low variance as significant (also a problem when the sample size is small). Quite evident in the Volcano plot. This is addressed by eBayes and SAM, which use the array wide standard deviation to adjust the gene/probe standard deviation. I don't know if i want to go to the trouble of reanalyzing this MA data using LIMMA. Maybe i should employ a simple fold-change filter?
- Media was calibrated to induce, siderophore biosynthesis. SidA (Afu2g07680) is first step in siderophore biosyntesis. These are not DE in p0.01 t tests, but is DE in 2h p 0.05 t test (and no others). Not sure why SidA expression was not captured? maybe it was not that highly expressed?
- Between array normalization appears not to be done. Does this affect any downstream analysis?
In the paper "SreA-mediated iron regulation in Aspergillus fumigatus" <pubmed> 18721228 </pubmed>, found that SreA is a main regulator for iron uptake proteins (siderophore etc). Using the list of SreA response genes from this paper (49 in total incl SidA), I crossed ref the list of DE genes in response to low iron for the 3 time points (p 0.05)
The list of SreA regulated genes are in ~/data/projects/a_fungus/2/ma_data/sreA_ko_study/sreA_reg_genes.pvals.
linux command used: cut -f 12 sig_probes_4h_p005.txt | sort | join - sreA_reg_locus_tags.txt | uniq | wc -l (sreA_reg_locus_tags.txt already sorted)
check overlap_* files for list of common genes.