Evaluating single-subject study methods for personal transcriptomic interpretations to advance precision medicine

BMC Medical Genomics

Table 1 DEG Methods for Single-Subject Studies and their previous validations

Method	Experimental Design	Distribution Assumptions	P-value	Validation of method for single subject inference in original methods publications
				Internal		External
				Simulation	Biological Replicates or Gold Standard	Translation to diagnosis, prognosis & treatment
edgeR [21]	r	NB	✓	✓^a	✗	✗
DESeq [22]	r	NB	✓	✓^a	✗	✗
DESeq2 [23]	r	NB	✓	✓^a	✗	✗
DEGseq [24]	r	B	✓	✓^a	✗	✗
NOISeq^a [15]	r/ss (as NOISeq-sim)	NP	✓	✓^a	±^b	✗
Mixture Model [5]	ss	MM	±^c	✓	✗	✗
iDEG [14]	ss	NB	✗	✓^a	✗	✗

NB Negative Binomial, B Binomial, NP Non-Parametric, MM Mixture Model, ss single-subject analytics, r analytics of between group of replicates, ✓ = completed, ± = partially addressed, ✗ = not addressed
^aNOISeq-Bio was used to construct the reference standard, while NOISeq-sim was used in the single-subject prediction sets
^bPartial validation conducted using qPCR with 400 genes with ~ 80% DEGs
^cMixture Model provides a posterior probability rather than a p-value, when FDR < 5% is indicated in the manuscript, it translates as a posterior probability > 95% for the mixture models

ISSN: 1755-8794