Although synonymous mutations do not alter the encoded amino acids, they may affect protein function by interfering with the regulation of RNA splicing or altering transcript splicing. Advances in next-generation sequencing technologies have put the exploration of synonymous mutations at the forefront of precision medicine, some tools have been specifically designed for predicting the deleterious synonymous mutations.

In this work, we present a computational model, usDSM, using random forest to detect the deleterious synonymous mutation. The results on the test datasets indicate that we can achieve superior performance in comparison with other state-of-the-art machine learning classifiers.

Job Start

usDSM supports prediction of synonymous mutations in the GRCh37/hg19 assembly of the human genome.
Insert the list of synonymous mutations using the tab separated values format chr, pos, id, ref, alt (maximum 10,000 mutations for 5 columns) Example

Learn more about output

usDSM, the pre-computed score is available on here.

usDSM, the training and test datasets are available on here.


You can view the tutorial on here.


Xi Tang, Tao Zhang, Na Cheng, Huadong Wang, Chun-Hou Zheng, Junfeng Xia, Tiejun Zhang, usDSM: a novel method for deleterious synonymous mutation prediction using undersampling scheme, Briefings in Bioinformatics, Volume 22, Issue 5, September 2021, bbab123, https://doi.org/10.1093/bib/bbab123.
If you have any problem with the website, please contact Dr. Tiejun Zhang: zhang_tiejun@gzhmu.edu.cn or Junfeng Xia: jfxia@ahu.edu.cn

Note: usDSM is intended for research purposes only. Do not use the results to make clinical decisions.