Can Seurat use pseudocounts from kallisto (SMART Seq 2) ?

[Achiraa]

I’m working with single-cell RNA-seq data processed using kallisto, and I have estimated counts (sometimes referred to as pseudocounts). These are not raw UMI counts but the expected transcript/gene counts generated by kallisto's EM algorithm. These are the files I have:
!Sample_data_processing = singlecell_rnaseq_gene_counts.tsv.gz (pseudocounts of each gene)
!Sample_data_processing = singlecell_rnaseq_transcript_counts.tsv.gz (pseudocounts of each transcript)
!Sample_data_processing = singlecell_rnaseq_gene_tpm.tsv.gz (transcripts per million (TPM) of each gene)
!Sample_data_processing = singlecell_rnaseq_transcript_tpm.tsv.gz (transcripts per million (TPM) of each transcript)

My goal is to use this data in Seurat for clustering and differential expression analysis.

Questions

1. Is it acceptable to use these pseudocounts as input to Seurat (e.g., for Smart-seq2-style data)?
2. Should I round these estimated counts before running SCTransform or log-normalization?
3. Are TPMs from kallisto also supported, or is that discouraged?

I'm aware that tools like tximport convert these to DE-friendly formats for edgeR and DESeq2, but it's unclear what Seurat prefers for full-length protocols.

Thank you in advance for the clarification!