Change default parameters of SCAPE

General

This page provides information regarding how to change the default parameters of SCAPE. Here we are dealing with parameters in the statistical model, i.e. apamix. We assume the inputs have been properly processed. Since SCAPE is written in pure python code, one needs to modify the default parameters in the source code.

If you are using standard 10x or Microwell scRNA-seq datasets (mouse and human) generated by standard Illumina sequencing, then the default parameters should work well.

Location of default parameters in the source code

The default parameters are specified in "apamix.py" under folder "apamix" line 30-48.

The explanation of relevant parameters are provided as follows.

n_max_apa

Maximum number of pA sites. Default value is 5.

n_min_apa

Minimum number of pA sites. Default value is 1.

PolyA length parameters

Empirical polyA length parameters
"LA_dis_arr" are distinct lengths of polyA lengths in the dataset
"pmf_LA_dis_arr" are the number of reads for each distinct polyA length

If empirical polyA length parameters are set to "None", the algorithm will use uniform distribution by sampling from Unif(min_LA, max_LA).
"min_LA" minimum polyA length
"max_LA" maximum polyA length

Insert size parameters

"mu_f" mean of insert size
"sigma_f" std of insert size

min_ws

Minimum isoform component weight. If the inferred weight of an isoform component is less than "min_ws", this isoform component will be removed from the final result.

min_pa_gap

Minimum distance between pA sites. pA sites will be merged into one if their distance is less than this value

max_beta

Maximum std of pA sites. The algorithm will limit the std of pA sites to be less than this value in the numeric optimization.