egaffo
commented
1 year ago You can use the BYPASS option; set it in the vars.py configuration file.
You may also want to set the LINMAPS parameter.
More in detail, when you already have linear mappings, first, you need to get the (linearly) unmapped reads. Extract them from the BAM using samtools if you did not save them as FASTQ files.
Then, you have two choices with CirComPara2: 1) compute only the circRNA expression abundance, or 2) use your BAMs to compute also the data required for the circular-to-linear proportion calculation.
In case 1), you just put the unmapped reads as input in the meta.csv file and set BYPASS = 'linear,linmap' in the vars.py file. With such a setting, CirComPara2 will skip the linear transcripts quantification and linear read alignment steps, performing the circRNA detection and quantification solely. I assume you are not interested in using CirComPara2 also to analyse traditional gene expression; otherwise, set BYPASS = 'linmap' (let’s call it case 1a).
In case 2), you need to set both BYPASS and LINMAPS parameters as follows: BYPASS = 'linear,linmap' (or just 'linmap' in case 1a); LINMAPS must list the BAM’s path for each sample (mind that you also need the .bai BAM index files in the same location of each BAM) and the format is as a Python dictionary.
For instance, once, I split my analysis into two runs, one for the “linear” and one for the “circular” parts.
For the ‘linear’ analysis, I set BYPASS = ‘circular’ (that's all).
Then, in the subsequent run, I analysed the ‘circular’ fraction and set:
a) meta.csv with the unmapped read FASTQ files obtained in the first run (i.e.
sample,file
CRR026003,/reads/Mmul_linmap/samples/CRR026003/processings/unmapped_reads/unmapped_1.fastq.gz
CRR026003,/reads/Mmul_linmap/samples/CRR026003/processings/unmapped_reads/unmapped_2.fastq.gzNote that Mmul_linmap is the directory of the first run of CirComPara2);
b) BYPASS = 'linear,linmap' in the vars.py file, and
c) LINMAPS = "{'CRR026003':'/reads/Mmul_linmap/samples/CRR026003/processings/hisat2_out/CRR026003_hisat2.bam'}” also in the vars.py file.
Your LINMAPS parameter will have 1000 entries.
If your reads are stranded, and you want circRNAs with strand info, you need to set the HISAT2_EXTRA_PARAMS (f.i. HISAT2_EXTRA_PARAMS = '--rna-strandness RF') as this parameter controls all strandness-related stuff in CirComPara2 even when linmaps are skipped (I should rename this parameter).
Moreover, if your unmapped reads were not available as paired-end, you might want to also set CIRC_MAPPING = 'SE'in the vars.py file because all your in put reads are considered as single-ended.
Remember you can perform a “dryrun” with CirComPara2's -n option (in the command line) to check that the files can be read by CirComPara2 before actually running the analysis.
However, I've never tried using BAMs generated with anything other than HISAT2.
Hope it helps.
lovebaboon1989
This is the what the bam file I used looks like.
Hi there, I'd like to try using circompara2 to map circRNA reads in human genome for my RNAseq project. But since my project has over 1000 samples and we've already done the alignment work for mRNA by STAR aligner (which took lots of time), so I just wonder whether circompara2 may allow for the STAR-generated BAM file as a input? Because if we rerun the genome alignment pipeline using pair-end fastq files once again, that could cost extra a lot of time.. Thanks a lot and looking forward to your kind reply! Best regards, Weiqian Jiang