Error in running rseg (val_weight >0 assertion failed)

[dawe]

Hi all, I'm trying to run rseg like this

$ rseg-diff -c /usr/local/share/rseg/human-hg18-size.bed -d /usr/local/
share/rseg/deadzones-k36-hg18.bed -v -T -C -m 2 IP.bed Input.bed
[LOADING_DATA] chromosomes
[LOADING_DATA] dead zones
[LOADING_DATA] IP.bed
[LOADING_DATA] Input.bed
[LOADING_DATA] separating deserts
[Remove duplicate reads]
[Selecting bin size] use Hideaki's empirical method: Bin size = 10005
[preparing data] binning reads IP
[preparing data] binning reads Input
[preparing data] estimating parameters
rseg-diff: TwoStateScaleSplitResolveMixture.cpp:53: void
partition_values(const std::vector<double, std::allocator<double> >&,
const std::vector<double, std::allocator<double> >&, const
std::vector<double, std::allocator<double> >&, const
std::vector<double, std::allocator<double> >&, std::vector<double,
std::allocator<double> >&, std::vector<double, std::allocator<double>
>&, std::vector<double, std::allocator<double> >&, std::vector<double,
std::allocator<double> >&, std::vector<double, std::allocator<double>
>&): Assertion `val_weight > 0' failed.
Aborted

How can I get rid of the error?
thanks
d

[Song, Qiang]

Hi Davide,

First of all, the binning size 10005bp seems rather big. How many reads are there in your test sample? If you examine the read coverage on Genome Browser, can you see any enrichments where you think it should be?

As far as I can tell, the he failed Assertion `val_weight > 0' indicates that the read count in the test sample is equal to the corresponding window in the input sample. Can you manually check your data to rule out this possibility?

After ruling the above possibilities, I think we will have a better chance to figure out any other cause of the error.

Best,
Song Qiang

[Davide Cittaro]

Hi there,

On Jul 12, 2011, at 8:31 AM, Song, Qiang wrote:

Hi Davide,

First of all, the binning size 10005bp seems rather big. How many reads are there in your test sample? If you examine the read coverage on Genome Browser, can you see any enrichments where you think it should be?

The problem here is I'm totally blind on what's in the sample. The only thing I expect is a very spread and low signal (it is a "rare" histone marker), possibly very similar to the test. I've got almost 27M reads both in the IP and in the sample

As far as I can tell, the he failed Assertion `val_weight > 0' indicates that the read count in the test sample is equal to the corresponding window in the input sample. Can you manually check your data to rule out this possibility?

That is expected :-(

I get the same error when running rseg without test:

$ rseg -c /usr/local/share/rseg/human-hg18-size.bed -d /usr/local/share/rseg/deadzones-k36-hg18.bed -v -T -C IP.bed

[…]

rseg: TwoStateResolveMixture.cpp:42: void partition_values(const std::vector<double, std::allocator<double> >&, std::vector<double, std::allocator<double> >&, std::vector<double, std::allocator<double> >&): Assertion `val_weight > 0' failed.

Aborted

d

[Song, Qiang]

Hi Davide,

Thanks for bringing this problem to our attention. I need to look further into
the cause of the error and it may take a while.

Have you tried to manually set the bin size with the -b option, for example "-b 1000"?
What is the result?

Best,
Song Qiang

[Davide Cittaro]

On Jul 13, 2011, at 6:12 AM, Song, Qiang wrote:

Hi Davide,

Thanks for bringing this problem to our attention. I need to look further into
the cause of the error and it may take a while.

Have you tried to manually set the bin size with the -b option, for example "-b 1000"?
What is the result?

Same error :-( 

I believe I should consider this a result: the IP and the Input are not different at all!

d