MOTIVATION: Beyond well-studied microRNAs, noncanonical small non-coding RNAs (sncRNAs) derived from longer parental templates such as tRNAs, rRNAs, and Y RNAs, are emerging as important regulators in various biological processes and diseases. Yet, analyzing these noncanonical sncRNAs from sequencing data remains challenging due to the intrinsic sequence heterogeneity and highly noisy nature. Conventional strategies either sum up all sequencing reads mapped to a parental RNA, which sacrifices the resolution of single sncRNA species, or treat each unique RNA species/sequence independently, which faces substantial noise in low-replicate settings. RESULTS: Here, we introduce FUSION (Family-level Unique Small RNA Integration), a computational tool bridging these conventional approaches by first quantifying unique sncRNA species and then aggregating them into their respective parental RNA families. This family-level integration captures the contributions of individual sncRNA species while enhancing statistical power and robustness for differential abundance analysis. FUSION includes two modules: FUSION_ms, which reduces noise and amplifies signals for multiple-sample comparison to detect family-level abundance changes even with a small sample size, and FUSION_ps, which is powered by paired-sample analysis and optimized for "1-on-1" differential abundance analysis in single-case studies. Both modules are validated by cross-lab discoveries of dysregulated sncRNA families that could not be identified using conventional methods. In summary, FUSION provides a powerful framework for sncRNA sequencing data analysis, enhancing data interpretation and supporting small sample research. AVAILABILITY AND IMPLEMENTATION: FUSION is available at https://github.com/cozyrna/FUSION and archived at https://doi.org/10.5281/zenodo.16929712.