Paper / Creation and implications of a phenome-genome network

Creation and implications of a phenome-genome network
Atul J Butte & Isaac S Kohane
2006, Nature Biotechnology

▣ Background

- Comprehensively consider associations between components of phenotype, genotype and environment to identify genes that may govern phenotype and responses to the environment.

▣ Result

- Finding a network of relations between phenotypic, disease, environmental and experimental contexts as well as genes with differential expression associated with these concepts.

▣ Method

- 1. Seven fields of annotations representing the phenotype, environmental and experimental context from GEO samples, series and data sets are parsed and mapped to UMLS concepts.

• Downlaod GEO Data(http://www.ncbi.nlm.nih.gov/geo/)
• Extract seven fields of GEO data into DB tables(gds, gse, gsm, ...)
• Using MetaMap, MetaMap Automation Mapping Tool is been designed by Yonglae.

MetaMapApiTest_Yonglae.java

gds_title    GDS1531    Chito-oligomer effect    C1280500    -827    Effect    [qlco]
gds_title    GDS1531    Chito-oligomer effect    C2348382    -827    Effect    [qlco]
gds_title    GDS1531    on seedlings    C0242437    -1000    Seedlings    [plnt]
gds_title    GDS2674    20-hydroxyecdysone effect    C0013495    -734    20-Hydroxyecdysone    [horm, strd]
gds_title    GDS2674    20-hydroxyecdysone effect    C1280500    -827    Effect    [qlco]
gds_title    GDS2674    20-hydroxyecdysone effect    C0013495    -734    20-Hydroxyecdysone    [horm, strd]
gds_title    GDS2674    20-hydroxyecdysone effect    C2348382    -827    Effect    [qlco]
gds_title    GDS2674    on cultured larval organs    C0010453    -589    Culture    [idcn]
gds_title    GDS2674    on cultured larval organs    C0023047    -589    Larva    [euka]

- 2. GEO Platforms are manually related to NCBI Gene identifiers, allowing the same genes to be related across platforms.

• Using GRIP(http://grip.snubi.org)
• Using BLAST(http://blast.ncbi.nlm.nih.gov/Blast.cgi)
  

- 3. Gene expression measurements are rank normalized within each GEO sample, then averaged across each GEO series.

• Normalized Rank(http://people.revoledu.com/kardi/tutorial/Similarity/Normalized-Rank.html)

- 4. Mean expression measurements for each gene in each GEO data set were related to the concepts mapped from each GEO data set.

False-positives-and-the-qvalue.pdf

Gene ID	CUI	High/Low	p-value	q-value	MRN	MRN Not	Concept	Not Concept
1	C0026845	2	0.00314	0	0.322	0.7322	4	5
1	C0205448	1	0.016635	0.03	0.75225	0.388	4	5
1	C0596981	2	0.00314	0	0.322	0.7322	4	5
9	C0026845	2	0.001849	0.125	0.231833	0.448684	6	38
9	C0037585	1	0.01069	0.17	0.499889	0.398343	9	35