LeafViz Gene-level vizualization: Error: nrow(myclusters) > 0 is not TRUE about leafcutter HOT 11 CLOSED

(the fix requires a leafcutter reinstall because it changes code in make_gene_plot)

from leafcutter.

Hi, thanks for using Leafcutter. Sorry to hear about these problems. Just to check, can you confirm this:

1. you are using the most recent version of Leafcutter
2. you have reinstalled the R package and are using R >= 3.2
3. the data you are trying to visualise has all been processed with the same version of Leafcutter

These questions are useful for me:

• what species are you working with?
• can you confirm which annotation file you used?
• is the genome you aligned to is compatible with the annotation you're using? weird chromosome naming conventions are a common source of problems.

from leafcutter.

Hi Jack,
Thank you very much for getting back to me - these aren't major problems, just thought you may want to know about them.

1. I have run on two leafcutter installs and encounterd the gene-level plot synonym issue running on both, and the gene-level nrow issue on the IMac, which
   was downloaded on 08/05/2010, so should be the latest version of leafcutter.

MacBook Air - Mac OS Sierra
Leafcutter: download git clone https://github.com/davidaknowles/leafcutter 25/04/2018
R version: R 3.4.1 (2017-06-30) -- "Single Candle", Platform: x86_64-apple-darwin15.6.0 (64-bit)
Python 2.7.10
Samtools Version: 1.3 (using htslib 1.3)

IMac - Mac OS El Capitan
Leafcutter: download git clone https://github.com/davidaknowles/leafcutter 08/05/2018
R version 3.3.2 (2016-10-31) -- "Sincere Pumpkin Patch", Platform: x86_64-apple-darwin13.4.0 (64-bit)
Python 2.7.14_2
Samtools Version: 1.3 (using htslib 1.3)

2. I haven't reinstalled R as:
   leafcutter_ds.R ran without any warnings or errors.
   ds_plots.R ran with 11 warnings, which I haven't viewed, as I have just run from the command line.
   prepare_results.R ran without any warnings or errors.

3. The data I am trying to visualize has all be processed with the same version of Leafcutter, most recently on IMac 10/05/2018.

4. I am using Leafcutter downloaded on 08/05/2018, run 10/05/2018.

5. I am working on homo sapien, using gencode_hg19 annotation sourced with: ./download_human_annotation_codes.sh

6. I have aligned with genecode_hg19 reference sourced from sanger [07/02/2018]
   curl -O ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_human/release_19/GRCh37.p13.genome.fa.gz
   curl -O ftp://ftp.sanger.ac.uk/pub/gencode/Gencode_human/release_19/gencode.v19.chr_patch_hapl_scaff.annotation.gtf.gz

Building star index as per STARmanual.pdf Section 2, 2.1. Basic options p4

STAR
--runThreadN 24
--runMode genomeGenerate
--genomeDir $DATADIR
--genomeFastaFiles $DATADIR/GRCh37.p13.genome.fa
--sjdbGTFfile $DATADIR/gencode.v19.chr_patch_hapl_scaff.annotation.gtf
--sjdbOverhang 100

Mapping following your workflow, but with multiple fastq per sample, and passing in Read groups:

STAR
--genomeDir $STARINDEX
--twopassMode Basic
--outSAMstrandField intronMotif
--readFilesIn $FQ1LST $FQ2LST
--readFilesCommand zcat
--outSAMattrRGline $RGID1LST
--outSAMtype BAM Unsorted

Re: Gene-level plot - Synonym issue - Error: Aesthetics must be either length 1
I have only noticed this for LVRN / AQPEP so far, where these are both listed on Genecards as synonyms.
This plot fail is reproducible - I have also done a mapping with UCSC sourced hg19 reference and encountered the same issue.

Re: Gene-leve plot - Error: nrow(myclusters) > 0 is not TRUE
I only spotted this yesterday. The error was present, but not consistently reproducible for the same gene-plot using the same dataset (genecode ref mapped),
or a second dataset (ucsc ref mapped), so appeared session related?

The genes for which this arrose: TNS1 and CIDEC.

Hope this helps, and let me know if you need any more info, but I won't have access to the IMac install for the next few days.

I have a question or two regarding best practice recommendations for the handling of multi-mappers and PCR duplicates for leafcutter analysis, but I guess that would be better posted on your google forum?

Best wishes
Drevansa

from leafcutter.

Sorry - I closed this in error.

from leafcutter.

Hello,

I have done a complete reinstall and update of R (and mac dependencies, including clang, gfortran, xquartz, devtools, java etc), reinstalled Leafcutter using the latest download and the reported issue persists.

R version 3.5.0 (2018-04-23) -- "Joy in Playing" Platform: x86_64-apple-darwin15.6.0 (64-bit)
Gene level plotting, in the same session of LeafViz fails in an inconsistent manner for the same gene working with the same dataset (generated with the new install).
So I can happily plot a few genes without issue, then encounter a failure to plot, where the
"nrow(myclusters) > 0 is not TRUE" error is displayed in red in the plot field.
Then plot a different gene, and then return to the failed gene and plot successfully.

I've also encountered a server error when attempting to download plots that have generated successfully, where resuming a download (sometimes several times over) has ultimately allowed me to download.

Is the reported issue perhaps related to a server issue?

from leafcutter.

I also meet the problem like nrow(exons_chr) > 0 is not TRUE

from leafcutter.

@jackhump did you get anywhere with this?

from leafcutter.

No, it's a weird intermittent bug.
Most of the time nrow(exons_chr) > 0 is not TRUE appears due to a lack of annotation over a particular region or a mismatch between chromosome names.
But what @drevansa describes happens intermittently.
@drevansa sent me their .Rdata file to have a look at and I couldn't work out what's causing the problem. Sorry David.

from leafcutter.

Thanks for looking at it @jackhump .

Can one of you send me the .Rdata please?

from leafcutter.

done

from leafcutter.

You'll be SHOCKED to hear this was yet another chromosome naming bug (e.g. chr1 vs 1). Fixed now, just strips the leading 'chr' if it's there. It was random because we had a sample function that only looked at a subset of chromosome names so it would make a different decision if some other contigs (e.g. GL000220.1) got randomly selected.

from leafcutter.