FLAIR collapse 5'/3' UTR and splice isoform issue

[mckalem]

Hello FLAIR Team,

I am working on annotating the transcriptome of a fungal organism using Nanopore Direct RNA Sequencing. I have DRS where I enriched for full-length transcripts by using a 5' adapter ligation strategy. I am inputting filtered full-length reads to FLAIR2. My organism of interest has many single exon transcripts and many regions with antisense transcripts in packed genomic loci.

I struggle to get consistent 5' and 3' end isoforms, and FLAIR2 is missing some splice isoforms. I tried running FLAIR2 using less stringent parameters to get the most isoforms.

This is how I am invoking FLAIR2:

flair align -g /home/muratcankalem/genomes/G217BV3/HcG217Bv3_contig.fasta -r /home/muratcankalem/Histo_Isoform_Analsis/minimap2/fasta/merged_filtered_1_13_25.fasta \
--threads 32 \
--nvrna \
--quality 5 \
--output "/home/muratcankalem/Histo_Isoform_Analsis/flair2_H_Y_combined/H_Y_ALIGN"

flair correct -q /home/muratcankalem/Histo_Isoform_Analsis/flair2_H_Y_combined/H_Y_ALIGN.bed \
--gtf /home/muratcankalem/genomes/G217BV3/HcG217BV3_mRNA.gtf \
-g /home/muratcankalem/genomes/G217BV3/HcG217Bv3_contig.fasta \
--output /home/muratcankalem/Histo_Isoform_Analsis/flair2_H_Y_combined/H_Y_CORRECT \
--nvrna \
--threads 32 \
--ss_window 10 \
--print_check

#Collapse_C

flair collapse -q /home/muratcankalem/Histo_Isoform_Analsis/flair2_H_Y_combined/H_Y_CORRECT_all_corrected.bed \
-g /home/muratcankalem/genomes/G217BV3/HcG217Bv3_contig.fasta \
-r /home/muratcankalem/Histo_Isoform_Analsis/minimap2/fasta/merged_filtered_1_13_25.fasta \
--gtf /home/muratcankalem/genomes/G217BV3/HcG217BV3_mRNA.gtf \
--output /home/muratcankalem/Histo_Isoform_Analsis/flair2_H_Y_combined/collapse_1_15_25/H_Y_COLLAPSE_1_15_25_C \
--threads 32 \
--mm2_args=-G2000 \
--trust_ends \
--no_gtf_end_adjustment \
--generate_map \
--max_ends 5 \
--isoformtss \
--end_window 20 \
--support 2 \
--filter ginormous

I attached 2 examples of missing either 5'/3' UTR annotation or splice isoforms:

  I7I48_02682: I expect this transcript to have 5' and 3' UTR isoforms by looking at the reads, but FLAIR2 does not identify the correct TSS/TES for the isoforms.
            
  I7I48_04800: I expect this to have an isoform originating from the mis-splicing of intron 1, but FLAIR2 has not identified it.
            

I am unsure if there's a reason that these are not called as isoforms. There are many instances like these that I observe where 5' and 3' ends are not correctly called (happy to share my fasta files and flair outputs). How can I modify my FLAIR2 invocation to have the most inclusive list of isoforms (especially the 5' and 3' UTR isoforms)?

I'd appreciate any advice you might have for me.

Thank you,
-Murat

I7I48_04800 FLAIR_C.pdf
I7I48_02682 FLAIR_C.pdf

[diekhans]

Having the FASTA files to reproduce the problem would be very useful.. thanks

[diekhans]

data files to reproduce the problem are in /private/home/markd/pub/for-colette/issue-409.zip

[mckalem]

Hello - Providing short-read data solved the issue with missing splice isoforms. I still have problems identifying 5' and 3' UTR isoforms.