# Why is a PacBio read length larger than the aligned reference region?

I recently had some Iso-Seq sequencing done on my organism catfish on the new Sequel platform and got weird alignments for a size selected 4 Kilobase and up fraction after running the isoseq3 pipeline. When I investigated the exon intron junctions using data from (Pac Bio Iso-Seq, Illumina RNA-seq) against our private reference genome, I get PacBio Iso-Seq reads where the sequence length is greater than the seq length on the reference for my bam alignments:

Seq length: 6751

Seq length on ref: 5267

When blasting the sequence in NCBI it shows 5% similarity to the publicly-available genome. I'll attach a JBrowse image with the same RNA sequenced with Illumina paired RNA-seq. I also included a image and circled the pacbio contig and Illumina is above.

## Question 1:

Is this a problem with JBrowse?

## Question 2:

Does this mean that some DNA was sequenced with the catfish cDNA?

## Question 3:

Why is this aligned sequence's length longer than the length of the reference genome region to which it is aligned?

Thanks

### Q1

Probably not - this is probably just displaying the contents of the input (I presume a bam file) exactly as they are encoded.

### Q2

The splicing looks a little weird, but the exon-mapping pattern in the 5' part of the gene makes it seem like RNA rather than DNA. I think that the large intron is what makes you think that it is DNA, correct?

• Explanations for weird splicing
• The weird read is a rare splice variant but a true mRNA.
• The weird read is not completely matured and has an intron that has not yet been spliced. The IsoSeq library prep protocol picked this up somehow and it ended up being sequenced.

## Q3

• Explanations for why Seq length: 6751 and Seq length on ref: 5267
• There is probably a large amount of clipping on the 5' or 3' end of the read. This could be polyA or some other transcribed-but-not-yet-completely-matured seq. Or, the extra sequence could be sequencing error, PacBio adapters, or a chimeric molecule.

### things to try

How to check what is going on:

I'm guessing that you had to upload a bam file to JBrowse to make the IsoSeq track. Use the sequence name that you got from Jbrowse and run the following command to extract the alignment info from the bam file:

samtools view /path/to/your/isoseq/file.bam | grep 'weird_read_id'


Look at the cigar string to see what it says about 5' or 3' clipping and corroborate with the sequence data.

• you are right the first 3702 bases are soft clipped. – user1238097 Dec 13 '18 at 16:24
• I'm curious to know what the clipped sequence is once you figure it out! – conchoecia Dec 13 '18 at 16:53
• I am still a mystified why it is clipping. I am starting to notice that many of the reads that display this property look chimeric where the start site is seen twice in the middle of the sequence. I also note that this was the 4kb and up fraction. So it might be assembly error you mentioned earlier I am feeling. – user1238097 Dec 17 '18 at 22:47
• Thanks for following up. Please feel free to post another question about the assembly error if you'd like! – conchoecia Dec 17 '18 at 23:45