Skip to main content

Table 2 Summary of NN Studies Reviewed

From: Neural networks for genetic epidemiology: past, present, and future

Publication

Input

Output

Hidden Layer

 

Type

Coding

Type

Coding

Number Layers

Number Nodes

Bhat et al. 1999

Binary

0 = absence of allele

Binary

1/0/0 = unaffected

1

15

  

1 = presence of allele

 

0/1/0 = mildly affected

  
    

0/0/1 = severely affected

  

Bush et al 2005

Discrete

-1, -1 = 1/1 genotype

Binary

0 = unaffected

GP evolved

  

0, + 2 = 1/2 genotype

 

1 = affected

  
  

+1, -1 = 2/2 genotype

    

Costello et al. 2003

Dicrete

Varied

Binary

0 = unaffected

Multiple variations

    

1 = affected

  

Curtis et al. 2001

Discrete

0 = AA genotype

Binary

0 = unaffected

2

3

  

1 = AB genotype

 

1 = affected

  
  

2 = BB genotype

    

Curtis 2007

Discrete

0 = AA genotype

Binary

0 = unaffected

2

3

  

1 = AB genotype

 

1 = affected

  
  

2 = BB genotype

    

Giachino et al 2007

Discrete and Continuous

Categorical values of genotypes and clinical features

Binary

0 = unaffected

1

unknown

    

1 = affected

  

Li et al. 1999

Discrete

IBD sharing

Binary

0/1 = concordant or not

Multiple variations

  

+1= shared allele

 

0/1 = affected or unaffected

  
  

-1 = unshared allele

    
  

0 = uninformative

    

Lin et al 2006

Discrete

Categories of genotype combinations

Binary

0 = non-response

Multiple variations

    

1 = response

  

Lucek and Ott 1997

Binary

0 = absence of allele

Binary

4 nodes for each trait (20 total nodes)

1

70

  

1 = presence of allele

 

0 = quantitative trait off

  
    

1 = quantitative trait on

  

Lucek et al. 1998

Discrete

IBD sharing

Binary

+1,+1 = target output

1

√220

  

+1= shared allele

 

0, +1 = noise

  
  

-1 = unshared allele

    
  

0 = uninformative

    

Marinov and Weeks 2001

Discrete

IBD sharing

Binary

+1,+1 = target output

1

√220

  

+1= shared allele

 

0, +1 = noise

  
  

-1 = unshared allele

    
  

0 = uninformative

    

Matchenko-Shimko and Dube 2006

Discrete

Three combinations of possible allele combinations, transformed to a 0–1 range

Binary

0 = control

Multiple variations

    

1 = case

  

Motsinger et al (2006a)

Discrete

-1, -1 = 1/1 genotype

Binary

0 = unaffected

GP Evolved

  

0, + 2 = 1/2 genotype

 

1 = affected

  
  

+1, -1 = 2/2 genotype

    

Motsinger et al (2006b)

Discrete

-1, -1 = 1/1 genotype

Binary

0 = unaffected

GE Evolved

  

0, + 2 = 1/2 genotype

 

1 = affected

  
  

+1, -1 = 2/2 genotype

    

North et al 2003

Discrete

0 = AA genotype

Binary

0 = unaffected

Multiple Variations

  

1 = AB genotype

 

1 = affected

  
  

2 = BB genotype

    

Ott 2001

Discrete

-1, -1 = 1/1 genotype

Binary

0 = unaffected

NA

  

0, +2 = 1/2 genotype

 

1 = affected

  
  

+1, -1 = 2/2 genotype

    

Pankratz et al. 2001

Discrete

IBD sharing

Binary

1/1 = affected/affected

1

4

  

+1 = shared allele

 

0/1 = affected/unaffected

  
  

-1 = unshared allele

    
  

0 = uninformative

    

Penco et al 2005

Discrete

Categories of allele combinations at each genotype

Binary

0 = unaffected

Multiple variations, including and evolutionary process

    

1 = affected

  

Pociot et al. 2004

Discrete

Number of categories per sliding window

Binary

0 = unaffected

Multiple variations

    

1 = affected

  

Ritchie et al. 2003

Discrete

-1, -1 = 1/1 genotype

Binary

0 = unaffected

GP evolved

  

0, + 2 = 1/2 genotype

 

1 = affected

  
  

+1, -1 = 2/2 genotype

    

Saccone et al. 1999

Discrete

IBD sharing

Binary

1/1 = affected/affected

18 variations

  

+1= shared allele

 

0/1 = affected/unaffected

  
  

-1 = unshared allele

    
  

0 = uninformative

    

Serretti and Smeraldi 2004

Discrete

SERPR*l/l = 1

Binary

0 = nonresponse

1

7

  

SERPR*l/s = 2

 

1 = response

  
  

SERPR*s/s = 2

    
  

TPH*C/C = 1

    
  

TPH*C/A = 2

    
  

TPH*A/A = 2

    

Shoemaker et al. 2001

Varied

Varied

Binary

0 = unaffected

1

unknown

    

1 = affected

  

Tomita et al 2004

Discrete

Homozygous for major allele = (0.1, 0.1)

Binary

0.9 = case

Multiple variations

  

Heterozygous = (0.1, 0.9)

 

0.1 = control

  
  

Homozygous for minor allele = (0.9, 0.9)

    

Zandi et al. 2001

Contin.

Pedigree-specific NPL scores

Binary

1,1 = case pedigree

15 variations

    

1,0 = control pedigree

 Â