Getting Results

All protein prediction results are distinguished from other results of the same type by a params_id. You should think of the params_id as a shorthand for the exact invocation which led to a particular result. The intent is to guarantee that results are reproducible, identify sets of results which are comparable (i.e., have the same params_id), and distinguish results of a similar type but with different runs (e.g., different gap penalties). You should always specify a params_id when searching for results. The params table contains all of the parameters available in Unison; try select * from params.

For example, let's say you want to look up Pfam_fs 17.0 results for TNFB_HUMAN. You'll need the pseq_id as described above (it's pseq_id 98 and the params_id (select params_id from params where name='Pfam_fs 17.0').

First, you should ensure that the the analysis you seek has been run by consulting the run_history table: 

unison=> select * from run_history_v where pseq_id=98 order by params;
 pseq_id | params_id |      params       | origin_id |    origin    | pmodelset_id |    modelset    |      ran_on      | failed
---------+-----------+-------------------+-----------+--------------+--------------+----------------+------------------+--------
      98 |        19 | BIG-PI default    |           |              |              |                | 2005-02-06 01:50 | f
      98 |         4 | EMBOSS/antigenic  |           |              |              |                | 2003-10-08 00:08 | f
      98 |         5 | EMBOSS/sigcleave  |           |              |              |                | 2003-10-13 16:12 | f
      98 |        11 | Genome BLAT       |           |              |              |                | 2004-02-11 10:48 | f
      98 |        38 | PMAP 2006-03-20   |           |              |              |                | 2006-03-23 15:11 | f
      98 |        21 | Pfam 17.0 fs      |     15038 | Pfam 17.0 fs |              |                | 2005-03-26 12:17 | f
      98 |        33 | Pfam 19.0 fs      |     15038 | Pfam 19.0 fs |              |                | 2006-01-12 08:34 | f
      98 |        34 | Pfam 19.0 ls      |     15039 | Pfam 19.0 ls |              |                | 2006-01-13 21:30 | f
      98 |         1 | Prospect2 default |           |              |            5 | Death Domains  | 2004-02-25 15:15 | f
      98 |         1 | Prospect2 default |           |              |           10 | Prospect2 FSSP | 2004-02-20 08:03 | f
      98 |        17 | Psipred v2.45     |           |              |              |                | 2005-06-17 07:05 | f
      98 |        28 | SignalP 3.0 (euk) |           |              |              |                | 2005-11-03 16:40 | f
      98 |        29 | TMHMM 2.0c        |           |              |              |                | 2005-11-16 14:05 | f
      98 |        20 | protcomp default  |           |              |              |                | 2005-08-10 05:40 | f
      98 |        12 | regexp            |           |              |           19 | Bcl-2 domains  | 2005-11-16 11:26 | f
      98 |        12 | regexp            |           |              |           11 | ITxM           | 2006-04-13 04:47 | f
      98 |         9 | tmdetect default  |           |              |              |                | 2006-02-21 17:47 | f

To start, select the raw features for this pseq_id and params_id from pahmm table: 

unison=> select pseq_id,start,stop,pmodel_id,mstart,mstop,score,eval
         from pahmm where pseq_id=98 and params_id=21;
 pseq_id | start | stop | pmodel_id | mstart | mstop | score | eval
---------+-------+------+-----------+--------+-------+-------+-------
      98 |    19 |   43 |  13606812 |      1 |    26 |     4 |  0.22
      98 |    59 |  100 |  13612088 |     96 |   136 |     2 |     2
      98 |    97 |  114 |  13608971 |    186 |   199 |     0 |   6.1
      98 |   102 |  233 |  13612264 |      1 |   142 |   208 | 8e-61
      98 |   156 |  175 |  13609641 |    306 |   327 |    -1 |   9.5
(5 rows)
To make this useful, you need to map the pmodel_id to a Pfam model name, description, and other info. You could do the join yourself, but we've already provided a view which performs the join and provides other useful information as well, as shown below. 
unison=> select * from pahmm_v where pseq_id=98 and params_id=21;
 params_id | pseq_id | start | stop | mstart | mstop | ends | pct_cov | score | eval  | pmodel_id |    acc    |   name    |                             descr 
-----------+---------+-------+------+--------+-------+------+---------+-------+-------+-----------+-----------+-----------+----------------------------------------------------------------
        21 |      98 |   102 |  233 |      1 |   142 | []   |     100 |   208 | 8e-61 |  13612264 | PF00229.8 | TNF       | TNF(Tumour Necrosis Factor) family
        21 |      98 |    19 |   43 |      1 |    26 | [.   |       8 |     4 |  0.22 |  13606812 | PF06682.2 | DUF1183   | Protein of unknown function (DUF1183)
        21 |      98 |    59 |  100 |     96 |   136 | .]   |      30 |     2 |     2 |  13612088 | PF01387.7 | Synuclein | Synuclein
        21 |      98 |    97 |  114 |    186 |   199 | .]   |       7 |     0 |   6.1 |  13608971 | PF01001.7 | HCV_NS4b  | Hepatitis C virus non-structural protein NS4b
        21 |      98 |   156 |  175 |    306 |   327 | .]   |       7 |    -1 |   9.5 |  13609641 | PF08124.1 | Lyase_8_N | Polysaccharide lyase family 8, N terminal alpha-helical domain
(5 rows)
IMPORTANT! If Unison returns no precomputed results where you expected otherwise, it is possible that the algorithm has not been run on the query sequence. You should always check the run_history table (or run_history_v view) to determine which predictive methods have been run on the query sequence.
Unison's real power, however, is in the inverse problem: given some criteria, find matching sequences. Feature-based mining requires precomputed data and this is where Unison excels. 
unison=> select pseq_id,eval,best_annotation(pseq_id) from pahmm_v where params_id=21 and pmodel_id=13612264 and eval<1e-5 order by eval;

 pseq_id |  eval   |                                                             best_annotation
---------+---------+------------------------------------------------------------------------------------------------------------------------------------------
   96930 | 4.6e-61 | UniProt:TNFA_PIG (Tumor necrosis factor precursor (TNF-alpha) (Tumor necrosis factor ligand superfamily member 2) (TNF-a) (Cachectin))
 1232850 |   8e-61 | UniProt:TNFA_PANTR (Tumor necrosis factor precursor (TNF-alpha) (Tumor necrosis factor ligand superfamily member 2) (TNF-a) (Cachectin))
      98 |   8e-61 | UniProt:TNFA_HUMAN (Tumor necrosis factor precursor (TNF-alpha) (Tumor necrosis factor ligand superfamily member 2) (TNF-a) (Cachectin))
   96931 | 1.3e-60 | UniProt:TNFA_RABIT (Tumor necrosis factor precursor (TNF-alpha) (Tumor necrosis factor ligand superfamily member 2) (TNF-a) (Cachectin))
  953430 | 3.1e-58 | Refseq:NP_796345.1 (tumor necrosis factor (ligand) superfamily, member 15 [Mus musculus])
...
It is easy to restrict sequences to particular species or require other criteria.