Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning

Challenges and solutions

Yoonha Choi, Tiffany Ting Liu, Daniel G. Pankratz, Thomas V. Colby, Neil M. Barth, David A. Lynch, P. Sean Walsh, Ganesh Raghu, Giulia C. Kennedy, Jing Huang

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Background: We developed a classifier using RNA sequencing data that identifies the usual interstitial pneumonia (UIP) pattern for the diagnosis of idiopathic pulmonary fibrosis. We addressed significant challenges, including limited sample size, biological and technical sample heterogeneity, and reagent and assay batch effects. Results: We identified inter- and intra-patient heterogeneity, particularly within the non-UIP group. The models classified UIP on transbronchial biopsy samples with a receiver-operating characteristic area under the curve of ~0.9 in cross-validation. Using in silico mixed samples in training, we prospectively defined a decision boundary to optimize specificity at ≥85%. The penalized logistic regression model showed greater reproducibility across technical replicates and was chosen as the final model. The final model showed sensitivity of 70% and specificity of 88% in the test set. Conclusions: We demonstrated that the suggested methodologies appropriately addressed challenges of the sample size, disease heterogeneity and technical batch effects and developed a highly accurate and robust classifier leveraging RNAsequencing for the classification of UIP.

Original languageEnglish (US)
Article number101
JournalBMC Genomics
Volume19
DOIs
StatePublished - May 9 2018

Fingerprint

Idiopathic Pulmonary Fibrosis
RNA
Sample Size
Logistic Models
RNA Sequence Analysis
Interstitial Lung Diseases
ROC Curve
Computer Simulation
Area Under Curve
Biopsy
Sensitivity and Specificity
Machine Learning

ASJC Scopus subject areas

  • Biotechnology
  • Genetics

Cite this

Choi, Y., Liu, T. T., Pankratz, D. G., Colby, T. V., Barth, N. M., Lynch, D. A., ... Huang, J. (2018). Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning: Challenges and solutions. BMC Genomics, 19, [101]. https://doi.org/10.1186/s12864-018-4467-6

Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning : Challenges and solutions. / Choi, Yoonha; Liu, Tiffany Ting; Pankratz, Daniel G.; Colby, Thomas V.; Barth, Neil M.; Lynch, David A.; Walsh, P. Sean; Raghu, Ganesh; Kennedy, Giulia C.; Huang, Jing.

In: BMC Genomics, Vol. 19, 101, 09.05.2018.

Research output: Contribution to journalArticle

Choi, Y, Liu, TT, Pankratz, DG, Colby, TV, Barth, NM, Lynch, DA, Walsh, PS, Raghu, G, Kennedy, GC & Huang, J 2018, 'Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning: Challenges and solutions', BMC Genomics, vol. 19, 101. https://doi.org/10.1186/s12864-018-4467-6
Choi, Yoonha ; Liu, Tiffany Ting ; Pankratz, Daniel G. ; Colby, Thomas V. ; Barth, Neil M. ; Lynch, David A. ; Walsh, P. Sean ; Raghu, Ganesh ; Kennedy, Giulia C. ; Huang, Jing. / Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning : Challenges and solutions. In: BMC Genomics. 2018 ; Vol. 19.
@article{d52f9c8da4864926a843bc3d5a620066,
title = "Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning: Challenges and solutions",
abstract = "Background: We developed a classifier using RNA sequencing data that identifies the usual interstitial pneumonia (UIP) pattern for the diagnosis of idiopathic pulmonary fibrosis. We addressed significant challenges, including limited sample size, biological and technical sample heterogeneity, and reagent and assay batch effects. Results: We identified inter- and intra-patient heterogeneity, particularly within the non-UIP group. The models classified UIP on transbronchial biopsy samples with a receiver-operating characteristic area under the curve of ~0.9 in cross-validation. Using in silico mixed samples in training, we prospectively defined a decision boundary to optimize specificity at ≥85{\%}. The penalized logistic regression model showed greater reproducibility across technical replicates and was chosen as the final model. The final model showed sensitivity of 70{\%} and specificity of 88{\%} in the test set. Conclusions: We demonstrated that the suggested methodologies appropriately addressed challenges of the sample size, disease heterogeneity and technical batch effects and developed a highly accurate and robust classifier leveraging RNAsequencing for the classification of UIP.",
author = "Yoonha Choi and Liu, {Tiffany Ting} and Pankratz, {Daniel G.} and Colby, {Thomas V.} and Barth, {Neil M.} and Lynch, {David A.} and Walsh, {P. Sean} and Ganesh Raghu and Kennedy, {Giulia C.} and Jing Huang",
year = "2018",
month = "5",
day = "9",
doi = "10.1186/s12864-018-4467-6",
language = "English (US)",
volume = "19",
journal = "BMC Genomics",
issn = "1471-2164",
publisher = "BioMed Central",

}

TY - JOUR

T1 - Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning

T2 - Challenges and solutions

AU - Choi, Yoonha

AU - Liu, Tiffany Ting

AU - Pankratz, Daniel G.

AU - Colby, Thomas V.

AU - Barth, Neil M.

AU - Lynch, David A.

AU - Walsh, P. Sean

AU - Raghu, Ganesh

AU - Kennedy, Giulia C.

AU - Huang, Jing

PY - 2018/5/9

Y1 - 2018/5/9

N2 - Background: We developed a classifier using RNA sequencing data that identifies the usual interstitial pneumonia (UIP) pattern for the diagnosis of idiopathic pulmonary fibrosis. We addressed significant challenges, including limited sample size, biological and technical sample heterogeneity, and reagent and assay batch effects. Results: We identified inter- and intra-patient heterogeneity, particularly within the non-UIP group. The models classified UIP on transbronchial biopsy samples with a receiver-operating characteristic area under the curve of ~0.9 in cross-validation. Using in silico mixed samples in training, we prospectively defined a decision boundary to optimize specificity at ≥85%. The penalized logistic regression model showed greater reproducibility across technical replicates and was chosen as the final model. The final model showed sensitivity of 70% and specificity of 88% in the test set. Conclusions: We demonstrated that the suggested methodologies appropriately addressed challenges of the sample size, disease heterogeneity and technical batch effects and developed a highly accurate and robust classifier leveraging RNAsequencing for the classification of UIP.

AB - Background: We developed a classifier using RNA sequencing data that identifies the usual interstitial pneumonia (UIP) pattern for the diagnosis of idiopathic pulmonary fibrosis. We addressed significant challenges, including limited sample size, biological and technical sample heterogeneity, and reagent and assay batch effects. Results: We identified inter- and intra-patient heterogeneity, particularly within the non-UIP group. The models classified UIP on transbronchial biopsy samples with a receiver-operating characteristic area under the curve of ~0.9 in cross-validation. Using in silico mixed samples in training, we prospectively defined a decision boundary to optimize specificity at ≥85%. The penalized logistic regression model showed greater reproducibility across technical replicates and was chosen as the final model. The final model showed sensitivity of 70% and specificity of 88% in the test set. Conclusions: We demonstrated that the suggested methodologies appropriately addressed challenges of the sample size, disease heterogeneity and technical batch effects and developed a highly accurate and robust classifier leveraging RNAsequencing for the classification of UIP.

UR - http://www.scopus.com/inward/record.url?scp=85039797581&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85039797581&partnerID=8YFLogxK

U2 - 10.1186/s12864-018-4467-6

DO - 10.1186/s12864-018-4467-6

M3 - Article

VL - 19

JO - BMC Genomics

JF - BMC Genomics

SN - 1471-2164

M1 - 101

ER -