Skip to main content

Table 2 Performance of five miRNA prediction tools using two single miRNA knock-in and one miR-144/451 double knock-out experiments

From: miREM: an expectation-maximization approach for prioritizing miRNAs associated with gene-set

Input data

Predictions

 

Datasets

miRNA

DEG

miREM (based

ChemiRsc

GeneSet2MiRNAc

CORNAc

Sylamerd

 

involved

a

on EM)b

    

Cytoplasmic

hsa-miR-155

647

hsa-miR-155-5p

hsa-miR-

HSA-MIR-155

hsa-mir-155

has-miR-155

RNA-seq in

knock in

 

(1/160)

155-5p

(1/23)

(1/2)

 

U2OS cells

   

(1/118)

   

Cytoplasmic

hsa-miR-1

743

hsa-miR-1-3p

hsa-miR-1-

HSA-MIR-1 (1/6)

hsa-mir-1

has-miR-1

RNA-seq in

knock in

 

(3/9)

3p (1/65)

 

(2/4)

 

U2OS cells

       

Microarray in

miR-144/451

396

mmu-miR-144-

Not

mmu-mir-144

mmu-mir-144

mmu-miR-

mice

knock out

 

3p (1/2)

Applicablee

(3/3) *

(2/4) *

144

   

mmu-miR-451a

    
   

(2/2)

    
  1. aDEG list not applicable for Sylamer where a full gene list ranked by fold change was input
  2. bSettings: p-value threshold = 0.01, EM convergence parameter = 0.001, common mappings from 3 or more databases and nonconserved miRNAs not included
  3. cSettings: p-value threshold = 0.01 (for ChemiRs, the minimum number of databases is 5 out of 10)
  4. dRanking number / full prediction result number is not available in Sylamer
  5. eMouse databases are not provided
  6. *P-value threshold = 0.05 (no result with p-value threshold = 0.01)