Supplementary Table S1. Filtering steps of RNA-seq data.

Of the initial 78 samples, 8 were excluded from the analysis based on quality reports from the sequencing lab. Consequently, transcriptomic analysis was performed on the remaining 70 samples. To quantify gene expression level, Salmon (v1.10.3) was used. The following reference files were downloaded from GENCODE (https://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M36/) on January 24, 2025: gencode.vM36.pc_transcripts.fa.gz, gencode.vM36.lncRNA_transcripts.fa.gz, gencode.vM36.primary_assembly.annotation.gtf.gz, GRCm39.primary_assembly.genome.fa.gz.
To build the index, protein-coding and lncRNA transcript FASTA files were combined into a single reference file, and decoy sequences were generated using the corresponding genome FASTA. Indexing was performed using salmon index with default settings and the additional options -d decoys.txt-gencode.
Transcript quantification was subsequently carried out using salmon quant with the following parameters: --libType A --unmatedReads --useVBOpt --seqBias --gcBias --writeQualities --geneMap "GTF" --writeUnmappedNames. The output from Salmon represented expression levels in transcripts per million (TPM).
Following quantification, an additional six samples were excluded due to low mapping rates (< 35%) identified during quality control. To reduce noise in downstream analyses, only genes with TPM > 1 in at least three samples were retained.
Transcript-level quantifications were then aggregated to the gene level using the R package ‘tximport’ [26]. ENSEMBL gene IDs were mapped to gene names using the org.Mm.eg.db annotation package. To facilitate downstream processing, genes were sorted based on their geometric mean TPM values across all samples, from highest to lowest. Therefore, the resulting gene-level count matrix, comprising 19341 genes across 64 blastocyst samples (26 intact and 38 twin embryos), was used for subsequent analysis.