Does sequencing volume not affect the number of new transcripts identified by Bambu?

[xhuashen]

hello, developers
I have some question about the number of identifying novel transcript by bambu. the situation is that I apply the bambu to two individual samples, for the two sample, we get two tissue whole blood and pbmc, usually we considers pbmc is a part of whole blood, but I don't know the opinion is truth.
For sample A we have 10G base whole blood and 20G base pbmc, for sample B we have 10G base pbmc and 20G base whole blood. so we have 4 nanopore cDNA long read RNA seq fastq file. and we use bambu to identify novel transcript. The result have some out of my comprehension.
The two whole blood samples identify same number(339(10G base) vs 345(20G base)) of novel transcript, and the situation also happened in the two pbmc samples(234(20GB),241(10GB)). we use the default parameter for every sample. so , does sequencing volume not affect the number of new transcripts identified by Bambu?
In order to figure out the matter, I have implemented two down sample for 20G sample, and use bambu with default parameter. The result for two samples(20G) and corresponding down sample(10G)(extract half of reads randomly for 20G sample) is very out of my mind. The number of identifying novel transcripts is that 236(20G)vs234(down sample 10G), and 345(20G)vs 413(down sample 10G). For getting more comprehensive perspective, we have implemented the same pipeline to a public data of whole blood. the situation is 214(20G)vs 208(10G).
Looking forward to your reply, I really want to figure out the question from any aspect.