25
EXPORTS Cdd-id, Cdd-id-set, Cdd, Cdd-set, Cdd-tree, Cdd-tree-set;
25
EXPORTS Cdd-id, Cdd-id-set, Cdd, Cdd-set, Cdd-tree, Cdd-tree-set, Cdd-pref-nodes, Cdd-Project;
27
IMPORTS Date FROM NCBI-General
29
Biostruc-annot-set FROM MMDB
30
Bioseq FROM NCBI-Sequence
31
Seq-annot FROM NCBI-Sequence
32
Seq-entry FROM NCBI-Seqset
33
Org-ref FROM NCBI-Organism
34
Seq-id FROM NCBI-Seqloc
35
Seq-interval FROM NCBI-Seqloc
36
Seq-loc FROM NCBI-Seqloc
37
Seq-feat FROM NCBI-Seqfeat
38
Score-set FROM NCBI-Seqalign
27
IMPORTS Date FROM NCBI-General
29
Biostruc-annot-set FROM MMDB
30
Bioseq FROM NCBI-Sequence
31
Seq-annot FROM NCBI-Sequence
32
Seq-entry FROM NCBI-Seqset
33
Org-ref FROM NCBI-Organism
34
Seq-id FROM NCBI-Seqloc
35
Seq-interval FROM NCBI-Seqloc
36
Seq-loc FROM NCBI-Seqloc
37
Seq-feat FROM NCBI-Seqfeat
38
Score-set FROM NCBI-Seqalign
39
39
Cn3d-style-dictionary,
40
Cn3d-user-annotations FROM NCBI-Cn3d;
40
Cn3d-user-annotations FROM NCBI-Cn3d
41
PssmWithParameters FROM NCBI-ScoreMat;
43
-- dealing with lists of preferred tax-nodes
45
Cdd-org-ref ::= SEQUENCE {
47
active BOOLEAN DEFAULT TRUE,
48
parent-tax-id INTEGER OPTIONAL,
49
rank VisibleString OPTIONAL
51
Cdd-org-ref-set ::= SET OF Cdd-org-ref
53
Cdd-pref-node-descr ::= CHOICE {
55
description VisibleString
58
Cdd-pref-node-descr-set ::= SET OF Cdd-pref-node-descr
60
Cdd-pref-nodes ::= SEQUENCE {
61
preferred-nodes Cdd-org-ref-set,
62
model-organisms Cdd-org-ref-set OPTIONAL,
63
optional-nodes Cdd-org-ref-set OPTIONAL,
64
description Cdd-pref-node-descr-set OPTIONAL
42
67
-- Cdd's should not exist without a unique accession, but alternative id's may
43
68
-- be present as well. It is conceivable that a CD which is created as a merged
70
95
avglen INTEGER OPTIONAL -- average repeat length
99
Cdd-book-ref ::= SEQUENCE { -- record a link to Entrez Books
100
bookname VisibleString, -- abbreviated book title
101
textelement ENUMERATED { unassigned(0), -- type of element
102
section(1), -- a section or paragraph
103
figgrp(2), -- a figure or set of figures
105
chapter(4), -- a whole chapter
106
biblist(5), -- a lisf of references
107
box(6), -- an inserted box
108
glossary(7), -- glossary
109
appendix(8), -- appendix
111
elementid INTEGER, -- address of the text-element
112
subelementid INTEGER OPTIONAL -- exact address, used with section
73
115
-- The description of CDD's refers to the specific set of aligned sequences,
74
116
-- the region that is being aligned and the information contained in the
75
117
-- alignment. It may contain a lengthy comment
102
143
auto-updated(6), -- update finished, no
103
144
-- work necessary
104
145
claimed(7), -- is earmarked for curation
146
curated-complete(8),-- public curated member of a
105
148
other(255) }, -- for CD production?
106
149
update-date Date, -- Date of last version change
107
150
scrapbook SEQUENCE OF VisibleString, -- for storing curation notes
108
151
-- those won't make it into public
110
153
source-id Cdd-id-set, -- for linking back to source db
111
repeats Cdd-repeat -- to record repeat counts
154
repeats Cdd-repeat, -- to record repeat counts
155
old-root Cdd-id-set, -- to record short-term history
156
curation-status INTEGER { unassigned(0), -- to record curation status
157
prein (1), -- when CD is checked out from
158
ofc (2), -- the tracking database, for
159
iac (3), -- use within curation software
166
readonly-status INTEGER { unassigned(0), -- to record read-only status
167
readonly (1), -- when CD is checked out from
168
readwrite (2), -- the tracking database, for
169
other (255) }, -- use within curation software
170
book-ref Cdd-book-ref
114
173
Cdd-descr-set ::= SET OF Cdd-descr
219
280
Align-annot-set ::= SEQUENCE OF Align-annot
282
-- the Domain-parent records an evolutionary relationship which may not be
283
-- as simple as a classical parent-child relationship in a typical hierarchy,
284
-- i.e. where a CD is merely a specific subgroup ("child") of a more general
285
-- diverse alignment model ("parent"). A CD alignment model may be the result
286
-- of an ancient fusion event, combining two or more domains into a bigger unit
287
-- which has subsequently undergone a divergent evolutionary process similar to
288
-- what may have happened to a single "domain". A CD alignment model may
289
-- also reflect the result of a deletion event, where a specific subgroup
290
-- lacks part of a (set of) domain(s), but where the part present is found to
291
-- be highly similar to a putative "parent", with some added evidence for
292
-- an actual deletion, like from the distribution of truncated copies in phylogenetic
293
-- lineages. Deletion events which affect different parts of a set of
294
-- duplicated domain architectures may be indistinguishable from actual
295
-- fission events, which means that we may want to represent the latter as
296
-- deletions after duplication and do not need a special case for fissions.
298
Domain-parent ::= SEQUENCE {
300
parent-type INTEGER { classical (0), -- the classification of parent child relations
305
parentid Cdd-id, -- identify the section parent by accession
306
seqannot Seq-annot OPTIONAL -- contains the sequence alignment linking
307
-- CD alignment models, should align the
308
-- masters/representatives of each CD
312
-- record sequence trees generated by a suitable algorithm.
314
Sequence-tree ::= SEQUENCE {
315
cdAccession VisibleString OPTIONAL,
316
algorithm Algorithm-type,
317
isAnnotated BOOLEAN DEFAULT FALSE,
321
SeqTree-node ::= SEQUENCE {
322
isAnnotated BOOLEAN DEFAULT FALSE,
323
name VisibleString OPTIONAL,
324
distance REAL OPTIONAL,
326
children SEQUENCE OF SeqTree-node,
328
seqRange Seq-interval,
329
rowId INTEGER OPTIONAL
332
annotation Node-annotation OPTIONAL
335
Algorithm-type ::= SEQUENCE {
336
scoring-Scheme INTEGER { unassigned (0),
338
kimura-corrected (2),
340
aligned-score-ext (4),
341
aligned-score-filled (5),
345
clustering-Method INTEGER { unassigned (0),
347
neighbor-joining (2),
348
fast-minimum-evolution (3),
350
score-Matrix INTEGER { unassigned (0),
357
other (255) } OPTIONAL,
358
gapOpen INTEGER OPTIONAL,
359
gapExtend INTEGER OPTIONAL,
360
gapScaleFactor INTEGER OPTIONAL,
361
nTerminalExt INTEGER OPTIONAL,
362
cTerminalExt INTEGER OPTIONAL
365
Node-annotation ::= SEQUENCE {
366
presentInChildCD VisibleString OPTIONAL,
367
note VisibleString OPTIONAL
221
370
-- the Cdd is the basic ASN.1 object storing an annotated and curated set of
222
371
-- alignments (formulated as a set of pairwise master-slave alignments).
223
372
-- The alignment data are contained in Seq-annots, and a special type of
233
382
name VisibleString, -- a short name (can be the accession..)
234
383
id Cdd-id-set, -- this CD's Ids
235
384
description Cdd-descr-set OPTIONAL, -- status, references, etc.
236
seqannot SEQUENCE OF Seq-annot OPTIONAL, -- contains the CD alignment
237
features Biostruc-annot-set OPTIONAL, -- contains structure
239
-- or "core" definitions
240
sequences Seq-entry OPTIONAL, -- store as bioseq-set inside seq-entry
241
profile-range Seq-interval OPTIONAL, -- profile for this region only
385
seqannot SEQUENCE OF Seq-annot OPTIONAL, -- contains the CD alignment
386
features Biostruc-annot-set OPTIONAL, -- contains structure
388
-- or "core" definitions
389
sequences Seq-entry OPTIONAL, -- store as bioseq-set inside seq-entry
390
profile-range Seq-interval OPTIONAL, -- profile for this region only
242
391
-- also stores the Seq-id of the master
243
trunc-master Bioseq OPTIONAL, -- holds the truncated master, which
392
trunc-master Bioseq OPTIONAL, -- holds the truncated master, which
244
393
-- may be something like a consensus,
245
394
-- uses the same sequence coordinate
246
395
-- frame as the profile-range
247
posfreq Matrix OPTIONAL, -- relative residue frequencies
248
scoremat Matrix OPTIONAL, -- Position dependent score matrix
249
distance Triangle OPTIONAL, -- pairwise distances for all seqs.
250
parents Cdd-id-set OPTIONAL, -- this CD is the result of
251
-- a split (or merge)
252
children Cdd-id-set OPTIONAL, -- this CD has been split
253
siblings Cdd-id-set OPTIONAL, -- related CDs (common hits)
254
neighbors Cdd-id-set OPTIONAL, -- co-occurring CDs
255
pending SEQUENCE OF Update-align OPTIONAL, -- contains alignments from
256
-- update or "lower panel"
257
rejects SEQUENCE OF Reject-id OPTIONAL, -- SeqIds of rejected CD-
258
-- members, ignore in update
396
posfreq Matrix OPTIONAL, -- relative residue frequencies
397
scoremat Matrix OPTIONAL, -- Position dependent score matrix
398
distance Triangle OPTIONAL, -- pairwise distances for all seqs.
399
parent Cdd-id OPTIONAL, -- this CD is the result of a split
400
children Cdd-id-set OPTIONAL, -- this CD has been split, not used
401
siblings Cdd-id-set OPTIONAL, -- related CDs (common hits), not used
402
neighbors Cdd-id-set OPTIONAL, -- co-occurring CDs, not used
403
pending SEQUENCE OF Update-align OPTIONAL, -- contains alignments from
404
-- update or "lower panel"
405
rejects SEQUENCE OF Reject-id OPTIONAL, -- SeqIds of rejected CD-
406
-- members, ignore in update
259
407
master3d SET OF Seq-id OPTIONAL, -- record if CD has a 3D representative
260
alignannot Align-annot-set OPTIONAL, -- alignment annotation
261
style-dictionary Cn3d-style-dictionary OPTIONAL, -- record rendering styles
262
user-annotations Cn3d-user-annotations OPTIONAL -- user annotations in Cn3D
408
alignannot Align-annot-set OPTIONAL, -- alignment annotation
409
style-dictionary Cn3d-style-dictionary OPTIONAL, -- record rendering styles
410
user-annotations Cn3d-user-annotations OPTIONAL, -- user annotations in Cn3D
411
ancestors SEQUENCE OF Domain-parent OPTIONAL, -- list of parents
412
scoreparams PssmWithParameters OPTIONAL,
413
seqtree Sequence-tree OPTIONAL
265
416
Cdd-set ::= SET OF Cdd
419
-- Cdd projects store a set of CDs, typically related to each other
420
-- relationships would be specified using the ancestors fields in the
421
-- individual CD objects. For use with CD-Tree, a program to visualize
422
-- curated CD hierarchies and evidence for hierarchical family structures.
424
Cdd-Viewer-Rect ::= SEQUENCE {
425
top INTEGER, -- top coordinate
426
left INTEGER, -- left coordinate
427
width INTEGER, -- width
428
height INTEGER -- height
431
Cdd-Viewer ::= SEQUENCE {
432
ctrl INTEGER { -- viewer type
443
dart-selected-rows (10),
446
rect Cdd-Viewer-Rect OPTIONAL, -- viewer rectangle
447
accessions SEQUENCE OF VisibleString -- list of accessions associated with a viewer
450
Cdd-Script ::= SEQUENCE {
454
server-generated (2),
457
name VisibleString OPTIONAL, -- user assigned name/description
458
commands VisibleString -- actual script commands
462
-- cd colors are as: 0000FF for red, 00FF00 for green, FF0000 for blue
464
Cdd-Project ::= SEQUENCE {
465
cds SEQUENCE OF Cdd , -- cds
466
cdcolor SEQUENCE OF INTEGER, -- colors
467
viewers SEQUENCE OF Cdd-Viewer, -- Sequence viewers
468
log VisibleString, -- log
469
scripts SEQUENCE OF Cdd-Script OPTIONAL -- command scripts