iRefIndex/PBR Notes

Preliminary BIND PSI-XML parsing results.

Results

Importing MITAB output from the iRefIndex 5.1 build and a build done using the BIND PSI-XML data, then comparing the interaction details for BIND, the following distributions are observed:

Interactions

			iRefIndex 5.1
	(Total)		(9811)
			3818
PSI-XML	(8916)	2923	5993

To summarise:

• Common interactions: 5993
• Interactions only in iRefIndex 5.1: 3818
• Interactions only in BIND PSI-XML: 2923

Interactors

			iRefIndex 5.1
	(Total)		(6896)
			2511
PSI-XML	(5324)	939	4385

To summarise:

• Common interactors: 4385
• Interactors only in iRefIndex 5.1: 2511
• Interactors only in BIND PSI-XML: 939

Queries

Common interactions:

select count(distinct mitab_sources.rigid)
from i5_mitab_sources
inner join mitab_sources
  on i5_mitab_sources.rigid = mitab_sources.rigid
  and i5_mitab_sources.name = mitab_sources.name
where mitab_sources.name = 'bind';

Interactions only in iRefIndex 5.1:

select count(distinct rigid)
from i5_mitab_sources
where i5_mitab_sources.name = 'bind'
  and rigid not in (select rigid from mitab_sources where mitab_sources.name = 'bind');

Interactions only in BIND PSI-XML:

select count(distinct rigid)
from mitab_sources
where mitab_sources.name = 'bind'
  and rigid not in (select rigid from i5_mitab_sources where i5_mitab_sources.name = 'bind');

Common interactors:

select count(distinct X.uid)
from (select uidA as uid from i5_mitab_sources where name = 'bind' union select uidB as uid from i5_mitab_sources where name = 'bind') as X
inner join (select uidA as uid from mitab_sources where name = 'bind' union select uidB as uid from mitab_sources where name = 'bind') as Y
  on X.uid = Y.uid;

Interactors only in iRefIndex 5.1:

select count(distinct X.uid)
from (select uidA as uid from i5_mitab_sources where name = 'bind' union select uidB as uid from i5_mitab_sources where name = 'bind') as X
where X.uid not in (select uidA as uid from mitab_sources where name = 'bind' union select uidB as uid from mitab_sources where name = 'bind');

Interactors only in BIND PSI-XML:

select count(distinct X.uid)
from (select uidA as uid from mitab_sources where name = 'bind' union select uidB as uid from mitab_sources where name = 'bind') as X
where X.uid not in (select uidA as uid from i5_mitab_sources where name = 'bind' union select uidB as uid from i5_mitab_sources where name = 'bind');

Other Stuff

This is what I did for the unnecessary "new BIND vs. other sources" experiment...

Downloaded the following:

• BioGrid
• CORUM
• DIP
• HPRD
• Intact
• MINT
• MIPS (MPPI)
• OPHID

Modified the following:

• BioGrid - removed all but human data

Parsed all sources mentioned above.

Removed all non-9606 records.

Did ROG assignment and so on.

Made the reports.