Table 2.
Comparison of methodological approaches for analysis of gene expression data
| Dataset | Data generation | Gene ontology analysis | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GEO series | Publication | RNA Quantification Platform | Normalization | Adjustment for covariates | Definition expression | Expressed genes | Threshold FC | Threshold P | Test | Multiple testing | GO analysis | Threshold GO enrichment p | Test | Multiple testing |
| GSE2900 | Griffiths (2005) | Stanford University cDNA lymphochip two color microarray | Scaled to geometric mean of sample:reference signal ratio from all array features | NS | Signal threshold | 9869 | 2.5 (from median in > 4 samples) | 0.1 | Permutation | FDR | NA | NA | NA | NA |
| GSE5418 | Ockenhouse (2006) | Affymetrix U133A GeneChips | RMA | NS | NS | NS | No | 0.01 | SAM, t-test | FDR | Onto Express and Pathway Architect | 0.05 | NS | FDR |
| GSE15221 | Franklin (2009) and Sharma (2011) | Illumina Human WG-6 v2.0 | Cubic spline | NS | Signal threshold | NS | 1.7 | 0.01 | Paired t-test | FDR | Onto Express | Varying | NS | NS |
| GSE15221 | Hirako (2018) | Illumina Human WG-6 v2.0 | Cubic spline | NS | Signal threshold | NS | 1.5 | 0.01 | Permutation and t-test | FDR | DAVID, GSEA | 0.05 | Multiple | FDR |
| GSE26876 | de Carvalho (2011) | Affymetrix Human Gene 1.0 ST Array | RMA | NS | NS | NS | 1.5 | 0.05 | Student t-test | No | Ingenuity pathway analysis | NS | NS | NS |
| GSE33811 | Krupka (2012) | Affymetrix Human Gene 1.0 ST Array | RMA and Quantile | NS | Signal and variation threshold | 3110 | 2 | 0.05 | Paired t-test | No | Gene set enrichment analysis on selected GO terms | 0.01 | Paired t-test | FDR |
| GSE34404 | Idaghdour (2012) | Illumina Human HT-12 BeadChips | Quantile | Location, Sex, Hb, total cell counts (RBCs and WBCs) and ancestry | Signal and normality threshold | NS | 2 (for comparison) | 0.01 | ANOVA, ANCOVA | FDR | Gene set enrichment analysis on customized MsigDB database | 0.05 | NS | Bonferroni |
| GSE55843 | Jagannathan (2014) | Agilent Sure Print G3 Human Gene Expression 8 × 60K v2 gene expression microarrays | Quantile | NS | Signal threshold | NS | 2 | 0.05 | SAM | FDR | NA | NA | NA | NA |
| GSE53292 | Jaijyan (2015) | Illumina Genome Analyzer Iix 72SE | NS | NS | NS | NS | NS | 0.05 | t-test | No | GeneCodis3, Bingo 2.3 plugin (Cytoscape 2.8.3) | 0.05 | NS | NS |
| GSE50957 GSE52166 | Tran (2016) | Illumina HiSeq 2000 2 × 100 PE | TAMM | Batch, Sex, Age, Pre-infection baseline | Signal and variation threshold, removal Y chromosomes | NS | 1.5 | 0.05 | Limma | FDR | Ingenuity pathway analysis | 0.05 | Fisher exact test | FDR |
| GSE50957 GSE67184 | Vallejo (2018) | Illumina HiSeq 2000 2 × 100 PE | CPM, TPM | NS | Signal threshold | NS | NS | 0.05 | EdgeR | FDR | WGSEA, ToppGene, STRING | 0.05 | Multiple | FDR |
| GSE64338 | Subramaniam (2015) | Affymetrix Human Gene 1.0 ST Array | Nonlinear normalization based on Li-Wong methods | NS | NS | NS | 1.2 | 0.001 | Paired t-test | FDR | Ingenuity Pathway Analysis | 0.05 | NS | FDR |
| GSE64493 | Sullivan (2015) | Agilent Sure Print G3 Human Gene Expression 8 × 60K v2 gene expression microarrays | Quantile | NS | Signal threshold | NS | 1.5 | 0.03 | Limma | FDR | DAVID | 0.05 | NS | FDR |
| GSE67184 | Rojas-Penas (2015) | Illumina HiSeq 2500 2 × 100 PE | SNM | Location/time-point, subject (random effect) | Signal threshold | 6154 | No | 0.05 | NS | FDR | NA | NA | NA | NA |
| GSE67184 | Gardinassi (2018) | Illumina HiSeq 2500 2 × 100 PE | NS | NS | NS | NS | No | 0.05 | Limma, repeated measures ANOVA | FDR | GSEA on blood transcriptome modules (BTM, Li et al.) | 0.05 | permutation | FDR |
| GSE7586 | Muehlenbachs (2007) | Affymetrix U133 Plus 2.0 GeneChip | GC RMA | NS | NS | NS | 2.5 | 0.01 | t-test | No | NA | NA | NA | NA |
| GSE77122 | Tarawa (2017) | Agilent Sure Print G3 Human Gene Expression 8 × 60K gene expression microarrays | Each gene expression array dataset was normalized to the in silicon pool for the macrophages cultured with RBCs | NS | NS | NS | No | 0.05 | Paired t-test | No | DAVID | 0.05 | Fisher exact test | No |
| GSE93664 | Burl (2017) | Affymetrix Human Gene ST 2.0 gene array | RMA | NS | NS | NS | 2 | 0.05 | NS | No | STRING | 0.01 | NS | Corrected unspecified |
| GSE100562 | Quin (2017) | Illumina HiSeq 2500 2 × 50 PE | NS | NS | NS | NS | No | 0.05 | Limma | FDR | NA | NA | NA | NA |
| GSE1124 | Boldt (2019) | Affymetrix U133A + B GeneChips | RMA | NS | Signal threshold | NS | 1.9 | 0.004 | SAM | FDR | DAVID and Ingenuity Pathway Analysis | 0.05 | NS | NS |
| GSE114076 | Terkawi (2018) | Agilent Sure Print G3 Human Gene Expression 8 × 60K gene expression microarrays | Each gene expression array dataset was normalized to the in silicon pool for the neutrophils cultured with RBCs | NS | NS | NS | 2 | 0.01 | Limma | No | Genomatix GeneRanker, DAVID, NET-GE and Enricher | 0.05 | NS | Corrected unspecified |
| GSE97158 | Rothan (2018) | Illumina HiSeq 2500 2 × 51 PE | TMM | Blocking by subject, in two separate models interaction with cell count and time of parasitemia was added | Signal threshold | 16,473 | 1.5 | 0.05 | Limma | FDR | GSEA (camera) on blood transcriptome modules (BTM, Li et al.) | 0.05 | Fisher exact test | FDR |
| GSE65928 | Portugal (2015) | Affymetrix Human Gene ST 2.0 gene array | RMA | NS | NS | NS | NS | 0.05 | ANOVA | FDR | Ingenuity pathway analysis | NS | NS | NS |
| GSE72058 | Feintuch (2016) | Affymetrix Human Gene 1.0 ST array | RMA and Quantile | Peripheral parasitemia | NS | NS | No | 0.05 | t-test | No | GSEA, CateGOrizer and ingenuity pathway analysis | 0.2 and 0.06 | NS | FDR |
FDR false discovery rate, Hb haemoglobin, NA not available, NS not specified in publication, RBCs red blood cells, RMA Robust Multichip average, SNM supervised normalization of microarray, TMM trimmed mean of M-values, GEO Gene Expression Omnibus, GE gene ontology