Some StringTie transcripts have no strand, causing SQANTI3 to discard them

[Maria-Eduarda2024]

Hello,

I am running StringTie followed by SQANTI3, but I noticed that some transcripts generated by StringTie do not have strand information (“.”). When I provide this GTF to SQANTI3, these transcripts are discarded as unknown strand.

Here is an example of the pipeline and the commands I used:

StringTie command
stringtie MOCK1_Aligned.sortedByCoord.out.bam
-G /temporario2/11320602/STAR/NtaSR1_annotation_1.0.gtf
-p 4
--rf
-o stringtie_out/Mock1.gtf

SQANTI3 command
sqanti3_qc.py
--isoforms /temporario2/11320602/stringtie_out/merged.gtf
--refGTF /temporario2/11320602/STAR/NtaSR1_annotation_1.0.gtf
--refFasta /temporario2/11320602/STAR/NtaSR1_genome_1.0.fasta
--skipORF

SQANTI3 log (excerpt)
[INFO:2025-09-23 11:32:54,209] Write arguments to /temporario2/11320602/sqanti3_results/isoforms.qc_params.txt...
[INFO:2025-09-23 11:33:51,777] **** Correcting sequences
[WARNING:2025-09-23 11:33:52,826] Discarding unknown strand transcript: Chr01 StringTie transcript 25908083 25908404 1000 . . gene_id "MSTRG.860"; transcript_id "MSTRG.860.1";
[WARNING:2025-09-23 11:33:52,827] Discarding unknown strand transcript: Chr01 StringTie exon 25908083 25908404 1000 . . gene_id "MSTRG.860"; transcript_id "MSTRG.860.1"; exon_number "1";

Question

Why would StringTie generate transcripts without strand information, even when I specify --rf?

Does this usually indicate ambiguous alignments or insufficient coverage to infer strand?

Is there a recommended way to keep these transcripts in SQANTI3, or are they always discarded?

Any advice would be greatly appreciated!

Thanks in advance.