no results in altPSI(jcounts)

[peterinnes]

Hello,

I'm encountering an issue with jCounts().

I ran

jcounts <- jCounts(counts = gbcounts,
                 features = features,
                 minReadLength = 75,
                 libType="SE",
                 strandMode=2)

When checking the output, the alternative splice site PSI results dataframe altPSI(jcounts) is empty, with zero rows. My understanding is that this means there are no alternative splice site usage events being detected--is that correct? This lack of results seemingly causes the downstream jDUreport() function to fail at the step Running altPSI test as jDUreport() throws the following error message:

jDUreport(jcounts,
                 minAvgCounts = 5,
                 contrast = c(1,-1),
                 filterWithContrasted = TRUE,
                 runUniformityTest = FALSE,
                 maxPValForUniformityCheck = 0.2,
                 strongFilter = TRUE,
                 maxConditionsForDispersionEstimate = 24,
                 formula = NULL,
                 coef = NULL,
                 maxFDRForParticipation = 0.05,
                 useSubset = FALSE)

Running junctionsPJU test
Running junctionsPIR test
Running irPIR test
Running esPSI test
Running altPSI test
Error in `.rowNamesDF<-`(x, value = value) : invalid 'row.names' length 

I'm wondering if you have any idea why jCounts() is returning an empty altPSI() results dataframe?

Thank you!!
-Peter

[estepi]

Dear Peter, thanks for using ASpli

Can you tell me:

1.- Which genome are you using? Which is the output when you run binGenome function? Do you have alternative splicing?

2.- How did you do the aligment? Which tool? Against which reference?

Once we have this info, we will try to figure out what is going on

Thanks a lot

[peterinnes]

Hi Estefania, thanks for the quick reply!

1. I'm working with the common sunflower (Helianthus annuus) reference genome Ha412HO v2.0 and corresponding gff file HAN412_Eugene_curated_v1_1.gff3.gz from https://sunflowergenome.org/. The binGenome function returns a ~140Mb ASpliFeatures object. Looking through it though, it appears the annotations only contain one transcript/isoform per gene, so I guess there aren't any splicing events annotated--I can see how this would be an issue with some of the downstream analysis steps that rely on annotated AS bins.

2. I used STAR in two-pass mode, using the reference genome mentioned above. Also perhaps worth noting is that my sequence data comes from a close relative (H. petiolaris), so that may be a factor here as well.

Let me know what you think.

Thanks,
Peter

[estepi]

Hi Peter, thanks for your reply

According your info:

1.- Yes, I would check if annotation has different isoforms for the same gene. If not, we cant do much more with ASpli and you should work with another tool for quantifying isoforms...

2.- STAR is ok for dealing with junctions (splice aware alignments) if you align against the genome. If you are not aligning exactly against your specie, check at least % or reads aligned and check in BAM files if they have junctions

I'll be waiting for your reply

[peterinnes]

I confirmed the annotations have only one isoform per gene. I will pursue other tools for quantifying splicing like you suggest. Hopefully I get to use ASpli in the future--it seems like a great program. Thanks again for your help and advice!

-Peter

[estepi]

Hi Peter, thanks for your reply.

Yes, if you dont have annotated isoforms, ASpli cant estimate AS events.
Anyway, do you have exons and introns? If yes, you can have a look to count tables and estimate "putative" (not known) AS events.

good luck!