Dataset Information

REDBot: Natural language process methods for clinical copy number variation reporting in prenatal and products of conception diagnosis.

ABSTRACT: BACKGROUND:Current copy number variation (CNV) identification methods have rapidly become mature. However, the postdetection processes such as variant interpretation or reporting are inefficient. To overcome this situation, we developed REDBot as an automated software package for accurate and direct generation of clinical diagnostic reports for prenatal and products of conception (POC) samples. METHODS:We applied natural language process (NLP) methods for analyzing 30,235 in-house historical clinical reports through active learning, and then, developed clinical knowledge bases, evidence-based interpretation methods and reporting criteria to support the whole postdetection pipeline. RESULTS:Of the 30,235 reports, we obtained 37,175 CNV-paragraph pairs. For these pairs, the active learning approaches achieved a 0.9466 average F1-score in sentence classification. The overall accuracy for variant classification was 95.7%, 95.2%, and 100.0% in retrospective, prospective, and clinical utility experiments, respectively. CONCLUSION:By integrating NLP methods in CNVs postdetection pipeline, REDBot is a robust and rapid tool with clinical utility for prenatal and POC diagnosis.

SUBMITTER: Liu M

PROVIDER: S-EPMC7667294 | biostudies-literature | 2020 Nov

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

REDBot: Natural language process methods for clinical copy number variation reporting in prenatal and products of conception diagnosis.

Liu Mengmeng M Zhong Yunshan Y Liu Hongqian H Liang Desheng D Liu Erhong E Zhang Yu Y Tian Feng F Liang Qiaowei Q Cram David S DS Wang Hua H Wu Lingqian L Yu Fuli F

Molecular genetics & genomic medicine 20200922 11

<h4>Background</h4>Current copy number variation (CNV) identification methods have rapidly become mature. However, the postdetection processes such as variant interpretation or reporting are inefficient. To overcome this situation, we developed REDBot as an automated software package for accurate and direct generation of clinical diagnostic reports for prenatal and products of conception (POC) samples.<h4>Methods</h4>We applied natural language process (NLP) methods for analyzing 30,235 in-house ...[more]

PMID: 32961042

Similar Datasets

Project description:BackgroundThe 8p23.1 duplication syndrome and copy number variation of the 8p23.1 defensin gene cluster are cytogenetically indistinguishable but distinct at the molecular level. To our knowledge, the 8p23.1 duplication syndrome has been described at prenatal diagnosis only once and we report our experience with four further apparent duplications ascertained at prenatal diagnosis.MethodsAdditional material at band 8p23.1 was detected using conventional G-banded cytogenetics in each case. Multiplex Ligation-dependent Probe Amplification (MLPA) or Fluorescence In Situ Hybridisation (FISH) were used depending on whether only DNA (Cases 1 and 4) or cytogenetic preparations (Cases 2 and 3) were available from the laboratory of origin. The extent of the duplication in Case 1 was retrospectively determined using array Comparative Genomic Hybridisation (array CGH).ResultsThree cases of 8p23.1 duplication syndrome were found (Cases 1 to 3). Two were de novo and continued to term and the third, a paternally transmitted duplication, was terminated because of a previous child with psychomotor delay and 8p23.1 duplication syndrome. Case 1 was ascertained with a hypoplastic left heart but the ventricular septal and interventricular defects, in Cases 2 and 3 respectively, were found after ascertainment for advanced maternal age. By contrast, case 4 was a maternally transmitted copy number variation of the defensin cluster with normal outcome.ConclusionsOur data underline the need to differentiate 8p23.1 duplications from copy number variation of the defensin cluster using FISH, MLPA or array CGH. Cardiac defects were ascertained by ultrasound in only one of the three duplication 8p23.1 pregnancies but were visible in two of the three at 21 to 22 weeks gestation. Our results provide further evidence that both deletion and duplication of the GATA4 transcription factor can give rise to a variety of conotruncal heart defects with variable penetrance and expressivity.

Project description:Copy number variations (CNVs) are gain and loss of DNA sequence of a genome. High throughput platforms such as microarrays and next generation sequencing technologies (NGS) have been applied for genome wide copy number losses. Although progress has been made in both approaches, the accuracy and consistency of CNV calling from the two platforms remain in dispute. In this study, we perform a deep analysis on copy number losses on 254 human DNA samples, which have both SNP microarray data and NGS data publicly available from Hapmap Project and 1000 Genomes Project respectively. We show that the copy number losses reported from Hapmap Project and 1000 Genome Project only have < 30% overlap, while these reports are required to have cross-platform (e.g. PCR, microarray and high-throughput sequencing) experimental supporting by their corresponding projects, even though state-of-art calling methods were employed. On the other hand, copy number losses are found directly from HapMap microarray data by an accurate algorithm, i.e. CNVhac, almost all of which have lower read mapping depth in NGS data; furthermore, 88% of which can be supported by the sequences with breakpoint in NGS data. Our results suggest the ability of microarray calling CNVs and the possible introduction of false negatives from the unessential requirement of the additional cross-platform supporting. The inconsistency of CNV reports from Hapmap Project and 1000 Genomes Project might result from the inadequate information containing in microarray data, the inconsistent detection criteria, or the filtration effect of cross-platform supporting. The statistical test on CNVs called from CNVhac show that the microarray data can offer reliable CNV reports, and majority of CNV candidates can be confirmed by raw sequences. Therefore, the CNV candidates given by a good caller could be highly reliable without cross-platform supporting, so additional experimental information should be applied in need instead of necessarily.

Dataset Information

REDBot: Natural language process methods for clinical copy number variation reporting in prenatal and products of conception diagnosis.

Publications

REDBot: Natural language process methods for clinical copy number variation reporting in prenatal and products of conception diagnosis.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets