High-resolution sample size enrichment of single-cell multi-modal low-throughput Patch-seq datasets