Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders
Abstract Background Accurate identification of Transcriptional Regulator binding locations is essential for analysis of genomic regions, including Cis Regulatory Elements. The customary NGS approaches, predominantly ChIP-Seq, can be obscured by data anomalies and biases which are difficult to detect...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2021-09-01
|
| Series: | BMC Bioinformatics |
| Subjects: | |
| Online Access: | https://doi.org/10.1186/s12859-021-04359-2 |
| _version_ | 1819098756147052544 |
|---|---|
| author | Quentin Ferré Jeanne Chèneby Denis Puthier Cécile Capponi Benoît Ballester |
| author_facet | Quentin Ferré Jeanne Chèneby Denis Puthier Cécile Capponi Benoît Ballester |
| author_sort | Quentin Ferré |
| collection | DOAJ |
| description | Abstract Background Accurate identification of Transcriptional Regulator binding locations is essential for analysis of genomic regions, including Cis Regulatory Elements. The customary NGS approaches, predominantly ChIP-Seq, can be obscured by data anomalies and biases which are difficult to detect without supervision. Results Here, we develop a method to leverage the usual combinations between many experimental series to mark such atypical peaks. We use deep learning to perform a lossy compression of the genomic regions’ representations with multiview convolutions. Using artificial data, we show that our method correctly identifies groups of correlating series and evaluates CRE according to group completeness. It is then applied to the ReMap database’s large volume of curated ChIP-seq data. We show that peaks lacking known biological correlators are singled out and less confirmed in real data. We propose normalization approaches useful in interpreting black-box models. Conclusion Our approach detects peaks that are less corroborated than average. It can be extended to other similar problems, and can be interpreted to identify correlation groups. It is implemented in an open-source tool called atyPeak. |
| first_indexed | 2024-12-22T00:36:02Z |
| format | Article |
| id | doaj.art-177889c5f407484fb6fe69b6f268bdec |
| institution | Directory Open Access Journal |
| issn | 1471-2105 |
| language | English |
| last_indexed | 2024-12-22T00:36:02Z |
| publishDate | 2021-09-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Bioinformatics |
| spelling | doaj.art-177889c5f407484fb6fe69b6f268bdec2022-12-21T18:44:49ZengBMCBMC Bioinformatics1471-21052021-09-0122112610.1186/s12859-021-04359-2Anomaly detection in genomic catalogues using unsupervised multi-view autoencodersQuentin Ferré0Jeanne Chèneby1Denis Puthier2Cécile Capponi3Benoît Ballester4INSERM, TAGC, Aix Marseille UniversityINSERM, TAGC, Aix Marseille UniversityINSERM, TAGC, Aix Marseille UniversityUniversité de Toulon, CNRS, LIS, Aix Marseille UniversityINSERM, TAGC, Aix Marseille UniversityAbstract Background Accurate identification of Transcriptional Regulator binding locations is essential for analysis of genomic regions, including Cis Regulatory Elements. The customary NGS approaches, predominantly ChIP-Seq, can be obscured by data anomalies and biases which are difficult to detect without supervision. Results Here, we develop a method to leverage the usual combinations between many experimental series to mark such atypical peaks. We use deep learning to perform a lossy compression of the genomic regions’ representations with multiview convolutions. Using artificial data, we show that our method correctly identifies groups of correlating series and evaluates CRE according to group completeness. It is then applied to the ReMap database’s large volume of curated ChIP-seq data. We show that peaks lacking known biological correlators are singled out and less confirmed in real data. We propose normalization approaches useful in interpreting black-box models. Conclusion Our approach detects peaks that are less corroborated than average. It can be extended to other similar problems, and can be interpreted to identify correlation groups. It is implemented in an open-source tool called atyPeak.https://doi.org/10.1186/s12859-021-04359-2Genomic assayAnomaly detectionCis regulatory elementUnsupervised curationConvolutional autoencoderChIP-seq peak quality |
| spellingShingle | Quentin Ferré Jeanne Chèneby Denis Puthier Cécile Capponi Benoît Ballester Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders BMC Bioinformatics Genomic assay Anomaly detection Cis regulatory element Unsupervised curation Convolutional autoencoder ChIP-seq peak quality |
| title | Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders |
| title_full | Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders |
| title_fullStr | Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders |
| title_full_unstemmed | Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders |
| title_short | Anomaly detection in genomic catalogues using unsupervised multi-view autoencoders |
| title_sort | anomaly detection in genomic catalogues using unsupervised multi view autoencoders |
| topic | Genomic assay Anomaly detection Cis regulatory element Unsupervised curation Convolutional autoencoder ChIP-seq peak quality |
| url | https://doi.org/10.1186/s12859-021-04359-2 |
| work_keys_str_mv | AT quentinferre anomalydetectioningenomiccataloguesusingunsupervisedmultiviewautoencoders AT jeannecheneby anomalydetectioningenomiccataloguesusingunsupervisedmultiviewautoencoders AT denisputhier anomalydetectioningenomiccataloguesusingunsupervisedmultiviewautoencoders AT cecilecapponi anomalydetectioningenomiccataloguesusingunsupervisedmultiviewautoencoders AT benoitballester anomalydetectioningenomiccataloguesusingunsupervisedmultiviewautoencoders |