tomazc / iCount

iCount, protein-RNA interaction analytics
http://icount.biolab.si
Other
23 stars 26 forks source link

iCount xlsites--many reads with second-start not falling on segmentation #209

Open ryanmarina opened 2 years ago

ryanmarina commented 2 years ago

I am attempting to determine cross-linking site with mapped reads, but certain libraries seem to give the same warning about several second starts not falling along segmentation borders when no custom segmentation file is provided (default):

Detecting cross-links... All records in BAM file: 310737 Reads not mapped: 0 Mapped reads records (hits): 310737 Hits ignored because of low MAPQ: 0 Records used for quantification: 310737 Records with invalid randomer info in header: 310737 Records with no randomer info: 0 Ten most frequent randomers: : 310737 There are 175499 reads with second-start not falling on segmentation. They are reported in file: /home/rmarina/scratch/iclip_data_download/beds/iCLIP_TDP-43_Brain_High_mm_beds/iCLIP_TDP-43_Brain_High_mm_skipped.bam Saved to BED file (uniquely mapped reads): /home/rmarina/scratch/iclip_data_download/beds/iCLIP_TDP-43_Brain_High_mm_beds/iCLIP_TDP-43_Brain_High_mm_unique.bed Saved to BED file (multi-mapped reads): /home/rmarina/scratch/iclip_data_download/beds/iCLIP_TDP-43_Brain_High_mm_beds/iCLIP_TDP-43_Brain_High_mm_multiple.bed

Can you explain what this means with no segmentation file provided, and if providing a custom segmentation reference will help recover some of these reads?