almost no genes detected

[wangjiawen2013]

Hi,
I followed the alevin-fry tutorial to process my customized feature-barcoding data, there are 19000 feature barcodes with 40 bp length each, the result are very weird, the gene expression quantified by alevin-fry is only 4% of that detected by umi-tools:

here is the statistics from featureDump.txt

CB CorrectedReads MappedReads DeduplicatedReads MappingRate DedupRate MeanByMax NumGenesExpressed NumGenesOverMean
ATAAGTGG 2234215 2188286 40832 0.9794429 0.018659353 0.0018865345 5228 914

The deduplicated reads is only 40832, but it's 1003469 when using umi-tools. What's the reason ?