Hi Flow team,
I’m working on an iiCLIP experiment with human iPSC-derived cortical neurons (WT vs mutation, -/+ stress treatment). While there are components that I need to optimize in the iiCLIP protocol itself, the current dataset that I have is showing a fair amount of antisense reads and I’m not sure why that is.
I uploaded multiplexed paired end read files and an annotation sheet for the demultiplexing pipeline via Flow to create my samples. I tried a variety of sequence combinations for the annotation sheet to try to ensure that I was using the appropriate orientation of the 3’ and 5’ barcodes using the updated L7 Illumina compatible sequences. I have 3 unique indexes in the 3’ barcodes (same 3’ index used for the same genotype) and each sample has a unique 5’ barcode. My annotation sheet is shown below. I got the 5’ bc sequences on a document from Dr. Ule, but the 3’ bc I used a flow documentation sheet to determine the orientation.
For the 3’ barcode, I tried a few different orientations based on the sheet you had shared. I’ve included an example below:
Using the above, I ran the CLIPseq protocol and found the antisense reads on IGV. I then did an upload using just the Read 1 fasta files and 5’ barcode for demultiplexing. This did seem to reduce the antisense reads, and it also seems like some of the reads that were antisense in the paired end data almost flipped to be sense reads - though sometimes this was slightly shifted left or right and not just a direct antisense-sense swap. I’ve included the comparison between the paired end upload (PE) and single end upload (SE) using the same samples looking at TARDBP. (WT_untreated_R1 = WT_R1).
I’m not sure if there’s something wrong with my annotation sheet or if there is some other type of technical contamination or bioinformatic adjustment I can make, but I’d love any insight and I’m happy to provide any additional information that might be helpful!
Hi there! I have also emailed Martina about this. I think we need to check the demultiplexing run, there is a lot of helpful information in the Ultraplex log that will tell you how many reads assign to each 5’ barcode then 3’ barcode, which will help us to understand if you are losing reads somewhere - due to having a 3’ barcode you will want to use both reads for demultiplexing (incase any of your read 1s do not read into the 3’barcode) but then only the first read for the clipseq pipeline (we have tested paired end alignment and find it doesn’t improve mapping). To clarify in the Flow annotation you give the sequences as they will be in the read itself. So your 3’ barcode is correct as far as I can tell! Feel free to share any demultiplexing or clipseq executions with me (my username is CharlotteC) and I can look into it further for you.
Hi! Thank you for your response. I just shared 4 demultiplexing and one CLIPseq executions with you. We reused the same 5’ barcodes across 4 replicates so I demultiplexed each separately, hence the four runs.
We did note that the fragments are on the shorter side so a portion are being lost due to size filters. This is something I am troubleshooting currently in lab to hopefully improve for future submissions. I also did a few runs that included unexpected combinations of 3’ and 5’ barcodes that could occur from lane-to-lane and gel-to-gel contamination during the visualization gel step which increased the overall read matches a bit, but of course didn’t effect any of the mapped reads from the paired end demultiplexing.
After the sample creation is there a way to run the clipseq pipeline with just the read 1s as opposed to both of the reads? I may have missed it but I didn’t see any of the modifiable parameters that were to use just the fastq1 file. Thank you so much for your help!
I’ll have a look at this for you. Just to quickly answer your last question, it’s a checkbox under the samplesheet table when you go to run the pipeline, but it will only appear if you uploaded the samples as paired end to begin with.
I’m sorry I can’t see the executions and work, Martina would be able to help you share them with me. Cheers!
You’re listed as a user under permissions now so I believe the sharing should be all set now!
Megan
