Oxford-Nanopore eBook - 17

Going Long
To overcome these challenges, Tang applied long-read nanopore
sequencing to enable unambiguous identification and analysis
of full-length RNA transcripts, using nine flow cells on the
high-throughput Oxford Nanopore PromethION platform.
"With short-read sequencing you are only able to look at
these events in isolation," she says. "With long-read sequencing, you get a view of the longer transcriptional context of
the aberrant splicing events and you can better predict their
functional consequences."
Tang's team used some of the patient samples from Wu's
2016 study to see how much more information they could
get from sequencing the full transcriptome of CLL samples
with and without the SF3B1 mutation. In all, the samples they
used comprised B-cells from three healthy donors, three CLL
patients with a wild-type SF3B1 and three CLL patients with
SF3B1K700E. Notably, the K700E hotspot mutation in the
HEAT-repeat domain of SF3B1 has been shown to be associated with altered splice sites at the 3' end of introns.
Tang's study produced 149 million cDNA reads that passed
QC and were passed through the group's Full-Length Alternative Isoform analysis of RNA (FLAIR) pipeline. Tang says this
pipeline leverages the full-length transcript sequencing data
afforded by the nanopore platform.
In this workflow, the raw sequencing reads are aligned to the
human reference genome prior to correction using annotated
splice junctions. The corrected reads are then grouped by their
splice junction chain before each group is collapsed to form
a consensus sequence for each individual transcript. Finally,
the raw sequencing reads are reassigned to the collapsed
isoforms. Isoforms that pass a given coverage threshold are
retained and used as a high-confidence transcript reference.
Alternative splicing patterns can be very complex, Tang points
out. First, these are cancer samples, which are notorious for
being heterogenous. In addition, the splicing factor mutations are also complex. The team found that when they ran
short-read alternative splicing callers on their data, the callers
would often miss or mis-call some of the splicing events they
were expecting.

More FLAIR
As a result, the team developed a new splicing caller,
FLAIR-diffSplice, which calls the four main types of alternative
splicing events (alternative 3'- and 5'- splicing, intron retention, and exon skipping events). Further, the caller identifies
which isoforms support the inclusion of the event, which
isoforms support the exclusion of the event, and with the

quantification of the isoform in each patient users can then
perform statistical tests to determine which splicing events
are differentially found between patient groups.
This caller identified the same alternative SF3B1 3'- splice site
choice as detected using an alternative short-read sequencing-based analysis, confirming the validity of the approach.
This pipeline, Tang says, "will correct and validate the splice
junctions in the reads." They were able to incorporate the
short-read sequences they had for these samples. Based on
these data, FLAIR built a set of isoforms that were representative of the reads.

With long-read sequencing, you get
a view of the longer transcriptional
context of the aberrant splicing
events and you can better predict
their functional consequences.

Focusing on intron retention, Tang says that, even though in
their study the nanopore sequencing reads were relatively
short (approximately 1 kb), nanopore sequencing reads made
intron retention "much more obvious."
Tang also notes that thanks to the facility of nanopore technology to sequence full-length transcripts, "isoform productivity can be more confidently assessed." The team defined
an unproductive isoform as those that have a premature
termination codon that is 55 nucleotides or more upstream of
the most 3' splice junction. Demonstrating the validity of their
nanopore sequencing-based productivity assessment, Tang
shared data showing complete concordance with previous
studies of the highly characterized isoforms of SRSF1.
Upon examining the productivity of transcripts from the
mutant SF3B1 cell line, the expression of unproductive
isoforms was decreased in comparison to the other cell types
tested. Tang has also shown that the down-regulated unproductive intron retention genes are associated with kinase
signaling pathways, which might support tumor proliferation.
Summarizing her work, Tang says that nanopore sequencing,
combined with the FLAIR analysis workflow, enables the study
of "differential isoform usage, coordinated splicing events, and
isoform productivity prediction." Her group is now working on
developing downstream analyses pipelines looking at productivity, differential splicing analysis and isoform usage. n
*Nanopore Community Meeting, hosted by Oxford Nanopore
Technologies; New York: December 5-6, 2019.

Watch Alison's full talk

https://nanoporetech.com/alisontang
clinicalomics.com

https://www.nanoporetech.com/alisontang http://www.clinicalomics.com

Oxford-Nanopore eBook

Table of Contents for the Digital Edition of Oxford-Nanopore eBook