Different N_REP parameters

[MaryAnnEbeling]

In the input_params.txt files in your tutorial, you use another parameter N_REP to define the number of replicates (it's not explained in Preparing Input Parameter File. In the case of your example, both RNA and protein data have two replicates. I got some data, where the number of replicates is different between RNA and protein. How can this be handled?
Thanks, Mary

[guoshou]

PECAplus requires paired protein-RNA time series data. Number of replicates in RNA and protein must be equal.

[hwchoi912]

One other way is to average the data over replicates and work with one consolidated time course data from both (N_REP=1) -- you will lose statistical power, but it's a quick solution to your situation.

Hyungwon

[MaryAnnEbeling]

Thanks for the feedback. We considered averaging the replicates as well. We were also thinking of just duplicating the missing replicates or making up some symmetric replicates based on the variance of related data assuming that we are working with pooled RNAseq samples.

I tried to run PECA-core with our data-sets and the different approaches above, but at the moment I had now a success.
As long as the transcript data files contain zero transcript numbers for a gene in one or more time-points, I get the following error:

nprot = 6188
impute...
peca_core_bin: ../include_dir/Eigen/src/Core/Redux.h:413: typename Eigen::internal::traits<T>::Scalar Eigen::DenseBase<Derived>::redux(const Func&) const [with BinaryOp = Eigen::internal::scalar_max_op<double, double>; Derived = Eigen::PartialReduxExpr<const Eigen::CwiseUnaryOp<Eigen::internal::scalar_abs_op<double>, const Eigen::Matrix<double, -1, -1> >, Eigen::internal::member_sum<double>, 0>; typename Eigen::internal::traits<T>::Scalar = double]: Assertion this->rows()>0 && this->cols()>0 && "you are using an empty matrix"' failed.
Aborted

In the output files x.txt, y.txt, I can see that zero's were replaced by NA, which might have caused the problem. Can those zero states not be handled in PECA?

I remove all genes from the data that contain at least one zero state over time, it is still not running and throws another error (doesn't matter if I use one replicate or three, see above):

nprot = 5250
impute... done
1000
terminate called after throwing an instance of 'std::runtime_error'
what(): fit
Aborted

I would be very happy if you could help me out with any tips or feedback. All the examples that you are providing are working in my installation, and I prepared all the data and the input-file according to that.

[hwchoi912]

I think the best way to troubleshoot the issue is for us to take a look at your data. Is there any chance you can send us your data after masking gene/protein names?

We are not seeking formal collaboration or anything like that, just wish to provide a solution for your problem. Feel free to email me at hwchoi@nus.edu.sg.

Hyungwon