The clinical sequencing of tumors is usually performed on formalin-fixed, paraffin-embedded (FFPE) samples and results in many sequencing errors. Most of these errors are detected in chimeric reads caused by single-strand DNA molecules with microhomology. Our filtering pipeline, MicroSEC, focuses on the uneven distribution of mutations in each read and removes the sequencing errors in FFPE samples without eliminating the true mutations that are also detected in fresh frozen samples.