Timeline for What is a sensible number of gene/observations to explain PCA variance?
Current License: CC BY-SA 4.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 30, 2020 at 11:52 | comment | added | Ecg | Thanks for the discussion from all of you. Just to clarify, my PCA shows three main clusters which coincide with my "conditions", thefeore my PCA would be a visual supporting information that the genes are expressed differently between these three clusters. So I believe, like @Devon mentioned below that the number do not really matter, just by showing the top most variable genes should show enough evidence that these clusters are separated, because using all genes both PC1 and 2 already account for a reasonable "big" percentage of variance (48%). | |
| Nov 30, 2020 at 11:39 | vote | accept | Ecg | ||
| Nov 30, 2020 at 8:34 | answer | added | Devon Ryan | timeline score: 3 | |
| Nov 29, 2020 at 23:31 | comment | added | Maximilian Press | This is not really what PCA is for, if I understand the question correctly. You don't select observations to "explain" PCA variance, that is more a feature of inferential tools like e.g. ANOVA. I would prefer to use some other measure to discard genes that are simply not informative or measured accurately. For example, first throw out low-expressed or non-expressed genes that are in the noise range. If what you want is to find a minimal set of genes that discriminate between two "clusters", then there are much better tools to do that than PCA, e.g. LDA or something like that. | |
| Nov 29, 2020 at 22:45 | comment | added | StupidWolf | And it's not very clear from your question.. | |
| Nov 29, 2020 at 22:45 | comment | added | StupidWolf | If you are using DESeq2, you are performing a PCA on the top 1000 most variables genes and plotting that. Some of your DEGs might be among these genes but there might be other factors other than the condition of your interest. Agree with Bastian its a a matter of what you want to convey with this PCA | |
| Nov 29, 2020 at 21:23 | comment | added | Bastian Schiffthaler | Of course, you can subset however you want. You just have to adjust the message and be clear when communicating such a result. In this case, you have to specify that your PC1 explains 50% of the variance of all DE genes, or the top 1000 most variable genes.You however can no longer claim the PC explains 50% of the dataset. | |
| Nov 29, 2020 at 21:02 | comment | added | M__♦ | Hi, yes I'd guessed this. The more data the fuzzier the signal. There's possibly a way foward, its not to reduce the amount of data .. in 'big data' its not cool. I can't remember its implemented though @StupidWolf might know. Hmmj looking at 'all genes' trapping a limited' variance that is weird. | |
| Nov 29, 2020 at 19:31 | comment | added | Ecg | S the cluster IS HAPPENING, just the less genes I select, the more clear (more reproducibility between replicates etc), just I am unsure on how many genes I should select. | |
| Nov 29, 2020 at 19:29 | comment | added | M__♦ | By result I mean clear clusters within PCA | |
| Nov 29, 2020 at 18:24 | comment | added | Ecg | Yes, you are right, I am trying to understand where to set up the cut off to dump only relevant data into the PCA, however I do not understand when you say "when I get no result". | |
| Nov 29, 2020 at 18:20 | review | Low quality posts | |||
| Nov 30, 2020 at 7:35 | |||||
| Nov 29, 2020 at 18:17 | comment | added | M__♦ | Hi @Ecg ... this is a good question. What we need to understand is why not dump all the data into the PCA - it can certainly handle it. What you appear to be saying is 'how can I re-jig PCA to give a result when I get no result'. If you can edit your question to help here that would be great rather than a priori assume PC1 and PC2 comprise max resolution | |
| Nov 29, 2020 at 17:31 | review | First posts | |||
| Nov 29, 2020 at 18:17 | |||||
| Nov 29, 2020 at 17:28 | history | asked | Ecg | CC BY-SA 4.0 |