Error "CompressedIntegerList" - cnFreq

[guidmt]

Hi, I am getting this error when I try to use cnFreq:

Error in FUN(X[[i]], ...) : Subsetting operation on CompressedIntegerList object 'x' produces a result that is too big to be represented as a CompressedList object. Please try to coerce 'x' to a SimpleList object first (with 'as(x, "SimpleList")').

I know that this is a point already discuss: 339 312

My dataset consist of 331303 rows. I tried to run with a small number of rows (5000) but I got again error. I have tried also to install the last version of GenVisR (GenVisR_1.19.6) without resolve the problem. I am running the code on server. Do you have any suggestion to resolve this issue?

Many thanks in advance for the support,

Best,

Guidantonio

[zlskidmore]

The problem results from unequal copy number segment lengths, when this happens the function chops up the segments into bins so samples can be directly compared. Until I have time to rewrite this theres not much that can be done.

However one suggestion would be to try and equal out the segment coordinates where possible.

For example:

chr1 5 10 chr1 3 9

would be broken by the function into

chr1 3 5 chr1 5 9 chr1 9 10

if you can make small adjustments (where you couldn't see the difference based on the graph resolution anyway). It would make a big difference. So in the example above you would have 1 coordinate chr1 3 10.

Obviously you have to be carefull with this, it's a tradeoff between the accuracy and computational resources.

Let me know if that doesn't make sense

[guidmt]

Dear Zachary,

Thanks for your reply! I think that make sense. I will try to use this approach, I will tell you if it works!

Best,

Guidantonio

[zlskidmore]

Thanks Guidantonio,

Please do let me know, if it ends up helping I would likely put this in as a parameter at least as a hotfix!

[guidmt]

Dear Zachary,

I haven't so much time to try your solution, and I unfortunately I haven't got good results. I think that the reason is given by the fact that I need to improve the way to collapse the regions as you said. Do you known some straightforward approach to do this?

Some approaches that could resolve this problem are:

• reducing the number of samples from your data
• remove regions in which you are not interested (e.g. LOH)

I know that are not great suggestions but this is what I have tried to do, maybe could be useful for other users.

Best

Guidantonio