Is it stuck in an infinite loop?

Downstream analysis and statistics HAllA
1

Hi,

I have just started using HallA. I am trying it on a dataset with ~1000 rows and 80 columns, so not particularly large. It has now been running for 5 days and shows — Testing hypothesis level xxxxxxxx with 1 hypotheses. It has, so far, tested nearly a billion hypotheses. Is this normal? Is it stuck in an infinite loop?

Martin

2

Hi Martin, depending on the settings you’re using it may or may not be stuck. Did you change the association metric from the default Spearman? Some of the association metrics use permutation tests to evaluate significance and are consequently much slower.

If your installation of HAllA is up to date, you should have gotten a timing estimate after the similarity table was finished being computed. Did that never show up?

3

Hi Andrew,

Thanks for the reply.

I have changed the metric from Spearman to Pearson and that does work.

I am using HallA on the university cluster. I think that it is up to date but I don’t have any control over it if it isn’t. If I run it from terminal and then don’t look until it is doing the hypothesis testing, I don’t see the time estimate. At least, I haven’t made a note of it. As I remember, when I have seen it it says a few hours.

Thanks for your work on this tool.

Regards.

Marttin

image002.jpg

4

Pearson should be fast even for millions of pairwise comparisons, so it probably got stuck at some point. I would stop the job you have running and double-check the format of your inputs.

HAllA underwent a major re-structuring recently, so I highly recommend updating to the latest version if at all possible.

5

I have version 0.8.17. Is that up-to-date enough?

6

Unfortunately that’s the version from just before the major re-structuring. You can find installation instructions here: halla – The Huttenhower Lab

7

Yes, I’d seen the installation instructions. I’ll have to contact the IT department and see if they will update the installation.