LiBiNorm corrects systematic bias introduced by the library preparation protocol. The form of the bias depends on the type of protocol used, and different models are required:
Smart-seq, Smart-seq2, and similar: model BD
Poly-A tagging, Quartz-seq, and similar: model D
Random priming: model E
The default mode for LiBiNorm is that the bias compensation is based on model BD (Smart-seq). Alternative models can be selected instead using the -n M option where M is one of the 6 models A-E and BD
The default mode is that the model parameters are determined and used for a single model. The -N <filename> option calculates parameters for all 6 models (including A, B and C, which are included for completeness and are simplified versions of B, D, and BD), picks the best fitting model and outputs three results files that provide details of the capabilities of all the models.
The best fitting model does not necessarily correspond to the actual library preparation protocol that was used, and bias correction results need to be treated with caution.
Number of reads used
By default LiBiNorm uses a maximum of 100,000,000 reads for parameter estimation but this can be changed using -d N. Reducing the number of reads reduces the time taken to determine the parameters but can make the parameters less optimal.
Maximum gene length
By default LiBiNorm only uses data from genes/transcripts where the transcript length is less than 20,000 bases. This is because the presence of reads associated with a small number of outlier long genes can have an undue influence on the model determination. The -e N option allows this default to be overwritten.
Number of threads
By default LiBiNorm uses three threads for the parameter estimation but this can be changed with the -p N option.