Unison Tour: Example: ITIM Mining

Background

Immunoreceptor Tyrosine Inhibitory Motif-containing proteins are an important class of immune system regulators [see Vivier E, et al., Nat. Immunol. 9:503(2008)]. ITIMs are short sequences in the intracellular domain of immune receptors. Canonical ITIM-containing proteins possess an extracellular immunoglobulin domain and a transmembrane domain in addition to the intracellular ITIM. In this example, we'll write a query to mine for ITIM-containing proteins, modify the query to restrict our results to human sequences that have orthologs, and then gather data to build a hypothesis about alternative ITIM protein domain composition.

Mining for Canonical ITIM proteins

This simple biological model may be translated into a simple SQL command that searches for an extracellular Pfam immunoglobulin domain, a transmembrane domain (with TMHMM), and an intracellular ITIM. For convenience and reproducibility, it's frequently desireable to use a view to capture a search method, as shown here: 
CREATE OR REPLACE VIEW itim_search_v AS 
SELECT IG.pseq_id,IG.start as ig_start,IG.stop as ig_stop,IG.score,IG.eval,
       TM.start as tm_start,TM.stop as tm_stop,
       ITIM.start as itim_start,ITIM.stop as itim_stop
  FROM pahmm_current_pfam_v IG
  JOIN pftmhmm_tms_v TM ON IG.pseq_id=TM.pseq_id   AND IG.stop<TM.start
  JOIN pfregexp_v ITIM  ON TM.pseq_id=ITIM.pseq_id AND TM.stop<ITIM.start
 WHERE IG.name='ig' 
       AND IG.eval<1e-2
       AND ITIM.acc='MOD_TYR_ITIM';

The models in the above query are:

Identifying ITIM proteins is now a matter of selecting from that view: 

eg=> SELECT DISTINCT ON (pseq_id) *,best_annotation(pseq_id)
     FROM itim_search_v WHERE eval<1e-2 ORDER BY pseq_id;
 pseq_id | ig_start | ig_stop | score |  eval   | tm_start | tm_stop | itim_start | itim_stop |                                                                                                  
---------+----------+---------+-------+---------+----------+---------+------------+-----------+--------------------------------------------------------------------------------------------------
     234 |      157 |     215 |    24 | 0.00068 |      440 |     462 |        518 |       523 | UniProtKB/Swiss-Prot:SIGL5_HUMAN (RecName: Full=Sialic acid-binding Ig-like lectin 5; Short=Sigle
     254 |      158 |     213 |    36 | 1.9e-07 |      284 |     306 |        386 |       391 | UniProtKB/Swiss-Prot:VSIG4_HUMAN (RecName: Full=V-set and immunoglobulin domain-containing protei
     368 |       49 |     119 |    32 | 3.3e-06 |      450 |     472 |        632 |       637 | RefSeq:NP_068713.2 (AXL receptor tyrosine kinase isoform 1 [Homo sapiens].)
...

Mining for ITIMs with Orthology

Although the individual features in itim_search_v are non-specific, the conjunction of all three is much rarer. Nonetheless, we can improve our confidence in the human candidates by corroborating them with chimp or mouse orthologs from HomoloGene that have similar domain composition.

For speed, we'll first make a temporary materialized table of orthologs using the existing homologene_orthologs_v view. 

-- build a table of homologene ortholog pairs from
-- the existing view
CREATE TABLE homologene_orthologs_mv AS 
SELECT * FROM  homologene_orthologs_v ;
CREATE INDEX homologene_orthologs_mv_q  ON homologene_orthologs_mv (q_pseq_id );
CREATE INDEX homologene_orthologs_mv_qt ON homologene_orthologs_mv (q_tax_id );
CREATE INDEX homologene_orthologs_mv_t  ON homologene_orthologs_mv (t_pseq_id );
CREATE INDEX homologene_orthologs_mv_tt ON homologene_orthologs_mv (t_tax_id );
ANALYZE homologene_orthologs_mv ;

Now, we can use the materialized ortholog data and our canonical ITIM query from the previous section to define a new view to search for Human ITIM proteins and orthologs. 

CREATE OR REPLACE VIEW itim_orthologs_search_v AS
SELECT DISTINCT ON (OP.q_pseq_id,OP.t_pseq_id,OP.t_tax_id)
  OP.q_pseq_id as h_pseq_id,
  HI.ig_start,HI.ig_stop,HI.score,HI.eval,HI.tm_start,HI.tm_stop,HI.itim_start,HI.itim_stop,
  OP.t_pseq_id as os_pseq_id,OP.t_tax_id as os_tax_id,tax_id2gs(OP.t_tax_id) as "os_gs"

  -- OP: HomoloGene ortholog pairs
  FROM homologene_orthologs_mv OP

  -- HI & HA: Human ITIM and Human Annotation
  JOIN itim_search_v HI      ON OP.q_pseq_id=HI.pseq_id 

  -- OI & OA: Ortholog ITIM and Human Annotation
  LEFT JOIN itim_search_v OI      ON OP.t_pseq_id=OI.pseq_id 

  WHERE OP.q_tax_id=gs2tax_id('HUMAN')
  ORDER BY  OP.q_pseq_id,OP.t_pseq_id,OP.t_tax_id
;

eg=> create table itim_ortho_search as select * from itim_orthologs_search_v where eval<1;
eg=> select * from itim_ortho_search ;
 h_pseq_id | ig_start | ig_stop | score |  eval   | tm_start | tm_stop | itim_start | itim_stop | os_pseq_id | os_tax_id | os_gs 
-----------+----------+---------+-------+---------+----------+---------+------------+-----------+------------+-----------+-------
       234 |      157 |     215 |    24 | 0.00068 |      440 |     462 |        518 |       523 |    1243401 |     10090 | MOUSE
       234 |      262 |     316 |    30 | 7.4e-06 |      440 |     462 |        518 |       523 |    5079208 |      9598 | PANTR
       234 |      262 |     316 |    30 | 7.4e-06 |      440 |     462 |        518 |       523 |    5605704 |     10116 | RAT
       234 |      157 |     215 |    24 | 0.00068 |      440 |     462 |        518 |       523 |    7203647 |      9615 | CANFA
       254 |      158 |     213 |    36 | 1.9e-07 |      284 |     306 |        386 |       391 |    5049591 |      9031 | CHICK
       254 |      158 |     213 |    36 | 1.9e-07 |      284 |     306 |        386 |       391 |    5081221 |      9598 | PANTR
[ 151 rows total ]

Exploring Alternative Domain Structures

eg=> CREATE OR REPLACE VIEW itim_ecds_v AS 
  SELECT count(*),ITIM.acc,ITIM.name,ITIM.descr
    FROM pahmm_current_pfam_v IG
    JOIN pftmhmm_tms_v TM ON IG.pseq_id=TM.pseq_id   AND IG.stop<TM.start
    JOIN pfregexp_v ITIM  ON TM.pseq_id=ITIM.pseq_id AND TM.stop<ITIM.start
   WHERE IG.eval<1e-2 AND ITIM.acc='MOD_TYR_ITIM' 
GROUP BY ITIM.acc,ITIM.name,ITIM.descr;

eg=> select * from itim_ecds_v ;

Beyond

The queries in this section may be modified to use other HMMs, threading or PSSMs to identify immunoglobulin domains, or to identify related Immunoreceptor Tyrosine Activating or Switch Motifs.