AmpliconNoise and Perseus are effective algorithms for removing noise from 454 pyrosequenced amplicons. AmpliconNoise improves upon PyroNoise by separately removing sequencing and PCR errors, while Perseus efficiently identifies chimeras using sequence abundances. These algorithms reduce OTU inflation and improve the accuracy of microbial diversity estimation. The study shows that chimeras significantly affect OTU counts, and Perseus achieves high sensitivity in detecting them. AmpliconNoise followed by Perseus effectively removes most noise, allowing accurate OTU construction. The algorithms are applicable to new sequencing technologies due to their use of Expectation-Maximization and supervised learning principles. The study evaluates the performance of these algorithms on various data sets, demonstrating their effectiveness in reducing noise and improving OTU accuracy. The results highlight the importance of noise removal in microbial diversity studies and the role of chimeras in OTU inflation. The algorithms are efficient and scalable, with AmpliconNoise being faster than PyroNoise. The study also shows that chimeras are a major source of OTU inflation, and their detection is crucial for accurate microbial diversity analysis. The results indicate that AmpliconNoise and Perseus are superior to other noise removal methods in terms of accuracy and efficiency.AmpliconNoise and Perseus are effective algorithms for removing noise from 454 pyrosequenced amplicons. AmpliconNoise improves upon PyroNoise by separately removing sequencing and PCR errors, while Perseus efficiently identifies chimeras using sequence abundances. These algorithms reduce OTU inflation and improve the accuracy of microbial diversity estimation. The study shows that chimeras significantly affect OTU counts, and Perseus achieves high sensitivity in detecting them. AmpliconNoise followed by Perseus effectively removes most noise, allowing accurate OTU construction. The algorithms are applicable to new sequencing technologies due to their use of Expectation-Maximization and supervised learning principles. The study evaluates the performance of these algorithms on various data sets, demonstrating their effectiveness in reducing noise and improving OTU accuracy. The results highlight the importance of noise removal in microbial diversity studies and the role of chimeras in OTU inflation. The algorithms are efficient and scalable, with AmpliconNoise being faster than PyroNoise. The study also shows that chimeras are a major source of OTU inflation, and their detection is crucial for accurate microbial diversity analysis. The results indicate that AmpliconNoise and Perseus are superior to other noise removal methods in terms of accuracy and efficiency.