I am trying to use the gene count table (.csv) to determine the differential gene expression using ballgown. As I filter my gene count table, I do not have any .ctab files. Is there any way to use ballgown for these gene count tables .csv files? For my plan of analysis, I MUST use ballgown ONLY.
Example of my gene count table:
gene_id Read Counts
FUN_000678 142 105 120 157 141 552 254 307 321 471
FUN_000679 114 143 63 151 96 210 169 142 116 201
Thanks in advance.
•
written
by
adnangeb05 •
Ballgown was designed to work with estimated transcript quantification values, not integer read counts as described in the documentation. If you MUST use ballgown, follow their workflow.
(I assume that your insistence on ballgown is a sign that you're aware that many people will recommend NOT using it and would instead suggest to use DESeq2, edgeR or limma-voom, which are all packages intended to be used with read counts)
I have to second that Ballgown should not be used. It always performs very poorly in benchmarks (like this one), have not been updated in years (according to the github) and is not even used by the authors themself (e.g. this resent large paper - where they use DESeq2).
Thank you for your explanation. I was trying to do a comparative analysis among the DEG tools after the manual filtering of my data set, as my gene count table is filtered, I was trying to use ballgown in the same ways like your recommended DESeq2, edgeR or limma. :)
You should not do that. The statistical model (the assumed distribution of the observations) are very different if you want to analyse counts or FPKM values. If you really want you could probably manually modify the Ballgown objet (overwriting the FPKM values with the count values) but if you are interested in the performance of FPKM value you should look into the limma-trend option instead.
Log in to answer this question.
