Discussion Points for 2nd Pseudogene Call

1
Discussion Points for 2nd Pseudogene Call

• Mark Gerstein
• 2005,09.22 1100 EST

2
Intersection of Pseudogenes from Three Groups
Original
42
45
Havana-Gencode 167 pseudogenes
35
21
86
Yale 184 pseudogenes
87
87
18
17
18
16
22
UCSC retrogenes 15 expressed (7-8 pseudogenes)
143 not expressed (all pseudogenes)
86 havana peudogenes overlap with any Yale
pseudogene and 87 Yale pseudogenes overlap with
any havana pseudogene (idem for retrogenes). This
is a global result maybe in some loci three
havana pseudogenes overlap with only one yale
pseudogene, but in other loci, several yale
pseudogenes overlap with one havana pseudogene.
Provided by France.
3
Intersection of Pseudogenes from 4 Groups Updated
52 (2)
Havana-Gencode 167 pseudogenes
14 (2)
16 (0)
Yale 164 pseudogenes
82 (34)
15 (1)
17 (7)
33 (1)
UCSC retrogenes 146 not expressed

• The numbers in parentheses are pseudogenes from
  GIS.
• All from http//pseudogene.org/ENCODE/cross-ref
• Pseudo-exons were merged to form pseudogenes and
  used for this comparison (now a pseudogene has
  only a single start and end)
• Strand information is ignored
• There are a total of 229 pseudogenes in the union

4
Intersection of Pseudogenes from 4 Groups
Non-processed Consensus
52 (2)
Havana-Gencode 167 pseudogenes
14 (2)
16 (0)
82 (34)
Yale 164 pseudogenes
15 (1)
17 (7)
33 (1)
UCSC retrogenes 146 not expressed
Roughly agreement now is 82 52 7 127 from
229 total What to do with 102?
GENCODE Processed GENCODE Non-Processed
Yale Processed 7 / 8 5 / 5
Yale Non-Processed 4 / 4 39 / 37
5
How to Pick Pseudogenes for RT-PCR?

• Start with the intersection 127
• Duplicated v processed how many of each? (21?)
• Rank Pseudogenes
• By likelihood to be transcribed according to
  ENCODE evidence
• ditag, then CAGE, then tiling array
• By their uniqueness in genome
• Good primers
• Non cross-hybridizing probes
• How to get a consistent rank?
• Who will do RT-PCR ?
• What coordinates to use ?
• (Ignore 1 processed pseudogene already being
  sequenced by GIS group.)

6
How to generate a consensus for remaining 102
pseudogenes?

• Stick with the intersection 127
• Develop a consistent criteria for identifying
  pseudogenes and uniformly apply to ENCODE
• E.g. protein matches with disablements found from
  a pipeline
• Ignores tricky cases flagged by manual annotation
• Do a simple union of UCSC, Havana Yale, giving
  229
• GIS is a subset of other 3
• Describe pseudogenes as being identified by
  multiple approaches and then explicitly flag each
  groups unique ones in final annotation
• Easy but perhaps biases stats
• Do a qualified union
• Allow each group to question particular
  pseudogenes in anothers set
• Send questions around and then have a call to
  sort out differences
• Need a way to arbitrate e.g. we could demand an
  obvious disablement
• We might learn something!
• How do we represent this in the browser in
  stats?

7
Once we have consensus, how to agree on
pseudogene boundaries?

• Keep unchanged each groups boundaries
• If pseudogenes overlap, take largest region
  (union) or smallest
• Develop a uniform criteria for assigning
  pseudogene boundaries and apply it to each of the
  pseudogenes in the consensus set
• Could just take each pseudogene in the consensus
  and have one group realign it against parent