Daniel-Liu-c0deb0t
commented
3 years ago Ok so UMICollapse was designed to work similarly to UMI-tools, where you would align to some reference first and then collapse reads at the same alignment coordinate. The fastq mode was a shortcut to not calculate the alignment and instead directly collapse based on the entire read sequence (including the UMI). Your use case of collapsing reads from a fastq directly based on the UMI in the header without alignment is not currently supported (although it shouldn't be too difficult of a feature to add; an alternative is to convert the fastq into a bam file and collapse that). Is there a specific reason why you are doing things this way? Usually, the number of duplicates would be overestimated if only the UMI is taken into account without the rest of the read sequence.
abracarambar
Hi Daniel, I am trying to run UMICollapse on a fastq file in which the UMI codes have already been extracted from the reads and incorporated in the read headers after a hash sign:
@CL100155237L2C001R001_12#ACAGAGGGTTC/1 AGTTGAACTCACCACGTGTCGGTA + DCCDDBCDDEDEDCDBE:DDBCBB @CL100155237L2C001R001_43#AAGTCTGTGTG/1 CAATGTGATTTCTGCCCAGTG + DBCCBECDBBDDDCDDECBBD @CL100155237L2C001R001_44#AGCCCATCGTC/1 TTGCCAAGGATGTTTTCATTAAT + CACEEDBBD@CCCCDCDCA@DDC
I have run the following command and it does collapse duplicates but ignores the barcodes in the read headers:
umicollapse fastq -i ${FASTQ_FILE} -o ${DEDUPED_FASTQ_FILE} --umi-sep "#" --tag Arguments [fastq, -i, ${FASTQ_FILE} -o, /${DEDUPED_FASTQ_FILE}, --umi-sep, #, --tag] Done reading input file into memory! Done with the first pass for tracking clusters! Number of input reads 44385266 Number of unique reads 4384633 Number of groups of reads 3587489 UMI collapsing finished in 618.248 seconds!
Any help would be greatly appreciated. M