All cand_circ disqualified in collapse

[cbohrson]

Hi,

As instructed by the readthedocs I ran CIRI-long collapse following the call command but I'm getting unexpected results.

I have inspected the isoforms generated by CIRI-long call in IGV after aligning them back to my reference genome (which in this case is a plasmid) and they match what I expect. However, collapse generates output like the following:

[Sun 2024-06-30 20:18:16] [INFO ] -------------- Collapse circular reads -------------
[Sun 2024-06-30 20:18:16] [WARNING] No annotation provided, entering 'De novo' mode
[Sun 2024-06-30 20:18:16] [INFO ] Step 1 - Clustering candidate circular reads
[Sun 2024-06-30 20:18:16] [INFO ] Circular reads clusters: 7
[Sun 2024-06-30 20:18:34] [100% ] [##################################################]
[Sun 2024-06-30 20:18:34] [INFO ] Corrected clusters: 0, 0/0/0/0 annotated/denovo/lariat/unknown
[Sun 2024-06-30 20:18:34] [INFO ] Step 2 - Calculating expression matrix
[Sun 2024-06-30 20:18:34] [INFO ] Final circRNAs: 0, isoforms: 0

In some cases corrected clusters will make it through but the number is much lower than the initial set.

Any possible reasons why all circular read clusters are being filtered? I have verified there is no clipping > 20 and the circRNAs are < 200,000 bp (I think these are filters from inspecting the code). I also modified collapse.py to not filter cluster size <= 1 and got the same results.