Standards and Guidelines for Validating NGS Bioinformatics Pipelines
A Joint Recommendation of the Association for Molecular Pathology and the College of American Pathologists
Un tema que preocupa a muchos clínicos y profesionales implicados en la NGS, es la validación de los algoritmos informáticos utilizados, ya que la repercusión en el ámbito clínico es de vital importancia.
Los pipelines bioinformáticos son un componente integral de la NGS. El procesamiento de los datos en bruto ( o raw sequence data) para detectar alteraciones genómicas, tienen un impacto significativo en el manejo de la enfermedad y atencion al paciente. Debido a la falta de orientación , actualmente existe un alto grado de variabilidad en cómo los miembros de la comunidad global de genética molecular y patología establecen y validan sus algoritmos que desarrollados de forma incorrecta y sin validación adecuada, pueden generar diagnósticos e incluso tratamientos erróneos y por tanto tener graves consecuencias para el paciente. Para abordar esta necesidad la Association of Molecular Pathology y la American Medical Informatics Association, han desarrollado un conjunto de 17 recomendaciones de consenso para la validación de algoritmos informáticos para la NGS aplicada a la clínica.
1 Clinical laboratories offering NGS-based testing should perform their own validation of the bioinformatics pipeline
2 A qualified medical professional with appropriate training in NGS interpretation and certification must oversee and be involved in the validation process
3 Validation must be performed only after completion of design, development, optimization, and familiarization of the bioinformatics pipeline and its components
4 Bioinformatics pipeline validation should closely emulate the real-world environment of the laboratory in which the test is performed
5 Validation should include all individual components of the bioinformatics pipeline used in the analysis, and each component must be reviewed and approved by an appropriately qualified medical molecular professional and the laboratory director
6 The design and implementation of the bioinformatics pipeline must ensure the security of identifiable patient information and be compliant with all applicable laws at the local, state, and national levels
7 Validation of the NGS bioinformatics pipeline must be appropriate and applicable for the intended clinical use, specimen, and variant types detected of the NGS test
8 Laboratories must ensure that the design, implementation, and validation of the bioinformatics pipeline are compliant with applicable laboratory accreditation standards and regulations
9 The bioinformatics pipeline is part of the test procedure, and its components and processes must be documented according to laboratory accreditation standards and regulations
10 The identity of the sample must be preserved throughout each step of the NGS bioinformatics pipeline with a minimum of four unique identifiers, including a unique location identifier within the content of each data file read and/or generated by the pipeline
11 Specific quality control and quality assurance parameters must be evaluated during validation and used to determine satisfactory performance of the bioinformatics pipeline
12 The methods used to alter or filter sequence reads at any point in the bioinformatics pipeline before interpretation must be validated to ensure that the data presented for interpretation accurately and reproducibly represent the sequence in the specimen, and full documentation of these methods must be kept as part of the test documentation according to laboratory accreditation standards and regulations
13 Laboratories must include specific measures to ensure that each data file generated in the bioinformatics pipeline maintains its integrity and provides alerts for or prevents the use of data files that have been altered in an unauthorized or unintended manner
14 In silico validation can be used to supplement the validation of the bioinformatics pipeline but shall not be used in lieu of end-to-end validation of the bioinformatics pipelines using human samples
15 Validation of the bioinformatics pipeline must include confirmation of a representative set of variants with highquality independent data; appropriate validation metrics by variant type should be reported
16 Clinical laboratories must ensure the accuracy of software-generated HGVS variant nomenclature and annotations and have an alert in place to indicate when the software-generated nomenclature and annotations need to be manually reviewed and/or corrected, and documentation of any corrections must be maintained
17 Supplemental validation is required whenever a significant change is made to any component of the bioinformatics pipeline
