Predicting Gene Functions from Text Using a CrossSpecies Approach

1
Predicting Gene Functions from Text Using a
Cross-Species Approach

• Emilia Stoica and Marti HearstSchool of
  InformationUniversity of California, Berkeley

Research Supported by NSF DBI-0317510 and a gift
from Genentech
2
Goal

• Annotate genes with functional information
  derived from journal articles.

3
Gene Ontology (GO)

• Gene Ontology (GO) controlled vocabulary for
  functional annotation
• 17,600 terms (circa July 2004)
• Organized into 3 distinct acyclic graphs
• molecular functions
• biological processes
• cellular locations
• More general terms are parents of less general
  terms
• development (GO0007275) is the parent of
  embryonic development (GO0001756)

4
Challenges

• GO tokens might not appear explicitly
• Example PubMed 10692450
• GO0008285 negative regulation of cell
  proliferation
• Occurs as inhibition of cell
  proliferation
• GO tokens might not occur contiguously
• Example PubMed 10734056,
• GO0007186
• G-protein coupled receptor protein signaling
  pathway
• Occurs as
• Results indicate that CCR1-mediated responses
  are regulated in the signaling pathway, by
  receptor phosphorylation at the level of
  receptor G/protein coupling CCR1 binds MIP-1
  alpha.

5
Challenges

• The simplest strategy (assigning GO codes to
  genes simply because the GO tokens occur near the
  gene) yields a large number of false positives.
• Issues
• The text does not contain evidence to support the
  annotation,
• The text contains evidence for the annotation,
  but the curator knows the gene to be involved in
  a function that is more general or more specific
  than the GO code matched in text.

6
Challenges

• GO contains hints about what kinds of evidence
  are required for annotation, e.g.
• The text should mention co-purification,
  co-immunoprecipitation experiments
• Requiring these evidence terms does not seem to
  improve algorithms.

7
Related Work

• Mainly in the context of BioCreative competition
  (2004)
• Chiang and Yu 2003, 2004
• Find phrase patterns commonly used in sentences
  describing gene functions
• (e.g., gene plays an important role in, gene
  is involved in)
• Final assignments made with a Naïve Bayes
  classifier
• Ray and Craven 2004, 2005
• Learn a statistical model for each GO code (which
  words are likely to co-occur in the paragraphs
  containing GO codes)
• Decide among candidates via a multinomial Naïve
  Bayes classifier
• Rice et al. 2004
• Train an SVM for each GO code.
• Target genes assigned best-scoring GO code.

8
Related Work, cont.

• Couto et al. 2004
• Determine if the information content of the
  matching GO terms is larger than for all the
  candidate GO terms.
• Verspoor et al. 2004
• Expand GO tokens with words that frequently
  co-occur in a training set use a categorizer
  that explores the structure of the Gene Ontology
  to find best hits.
• Ehler and Ruch 2004
• Treat each document as a query to be categorized
• Create a score based on a combination of pattern
  matching and TFIDF weighting
• Annotate gene with top-scoring GO codes.

9
Our Approach

• Two main contributions
• Use cross-species information (CSM)
• Check for biological (in) consistencies (CSC)

10
Cross-Species MatchMain Idea

• Use orthologous genes
• Genes of different species that have evolved
  directly from a common ancestor.
• Assumption
• Since there is an overlap between the genomes of
  the two species, their orthologs may share some
  functions, and consequently some GO codes
• Idea to predict GO codes for target genes in
  target species, use the GO codes assigned to
  their orthologous genes
• We use Mouse vs. Human genes

11
General procedure

• Analyze text at sentence level
• Eliminate stop words, punctuation characters and
  divide the text into tokens using space as
  delimiter
• Normalize and match different variations of gene
  names using the algorithm of Bhalotia et al.03
• For every sentence that contains the target gene
  
• A GO code is matched if the sentence contains a
  percentage of GO tokens larger than a threshold
  (0.75 for CSM and 1 for CSC)

12
Cross Species Match Algorithm

• CSM(g, a) For a target gene g, search in article
  a for only the GO codes annotated to its ortholog
  
• If at least 75 of the GO code terms are found in
  a sentence containing the gene name, the code is
  matched.
• Note we must eliminate annotations of orthologs
  marked with IEA and ISS codes to avoid circular
  references.

13
Cross-Species Correlation Main Idea

• Observation
• Since GO codes indicate gene function, it is
  logical for some to often co-occur in annotations
  and for others to rarely do so.
• Assumption
• If one GO code tends to occur in the orthologous
  genes annotations when another one does not,
  then assume the second is not a valid assignment
  for the target species
• Example
• If text seems to contain evidence for rRNA
  transcription (GO0009303) nucleolus (GO0005737)
  and extracellular (GO0005576), then
  extracellular is suspicious.
• The algorithm identifies the suspicious cases.

14
Cross-Species Correlation Algorithm

• For every pair of GO codes in the orthologous
  genes database, compute a X2 coefficient.
• N the total number of GO codes
• O11 of times the ortholog is annotated with
  both GO1 and GO2
• O12 of times the ortholog is annotated with
  GO1 but not GO2
• O21 of times the ortholog is annotated with
  GO2 but not GO1
• O12 of times the ortholog is not annotated
  with GO1 or GO2

X2
15
Cross-Species Correlation Algorithm

• M(g,a) GO codes matched in article a for gene g
• O(g) GO codes assigned to the ortholog of g
• o size of O(g), p percentage (0.2)
• For every potentially matching GO code GO1 in
  M(g,a)
• For every GO code GO2 in O(g)
• Count how often X2(GO1,GO2) is significant
• If this count is lt po then assume GO1 is not
  valid.
• Else assign GO1 to g

16
Information Flow
17
Evaluation using BioCreative

• Task 2.2
• Annotate 138 human genes with GO codes using 99
  full text articles
• For each annotation, provide the passage of text
  that the annotation was based upon.
• Annotations from participants were manually
  judged by human curators
• A prediction was considered perfect if the text
  passage
• contained the gene name, and
• provided evidence for annotating the gene with
  the GO code

18
Results on BioCreative

• Our research was conducted after the competition
  had past, so our annotations could not be judged
  by the same curators
• Used the perfect predictions
• (unfair to our system ignores relevant
  predictions we find that other systems do not)
• Our prediction is correct if it matches a perfect
  prediction (e.g., vhl is annotated with
  transcription (GO0006350) in PubMed 12169961
  vhl inhibits transcription elongation, mRNA
  stability and PKC activity)

19
BioCreative Results
20
Results on Larger Dataset

• A much larger test set has been made publicly
  available by Chiang and Yu.
• EBI human test set
• 4,410 genes
• 13,626 GO code annotations
• MGI mouse test set
• 2,188 genes
• 6,338 GO code annotations
• Note that Chiang and Yu used the same data for
  both training and testing.

21
Results on EBI Human and MGI datasets

• EBI human 4,410 genes and 5,714 abstracts
• MGI 2,188 genes and 1,947 abstracts

22
Conclusions and Future Work

• We propose an algorithm that annotates genes with
  GO codes using the information available from
  other species
• Experimental results on three datasets show that
  our algorithm consistently achieves higher
  F-measures than other solutions
• Future improvements to our algorithm
  - combine or use a
  voting scheme between the predictions our system
  makes and the predictions of a machine learning
  system
• - investigate how effective are other genes
  with sequences similar to the target gene (but
  not orthologous to the gene) for predicting the
  GO codes

23
Thank you!

• http//biotext.berkeley.edu

Research Supported by NSF DBI-0317510 and a gift
from Genentech
24
Example

• The marked accumulation of lipid droplets in
  LNCaP cells...is accompanied by an increase in
  phospholipid synthesis. The increase in PAP-2
  might be related to changes in lipid metabolism
  Since PAP-2 plays a pivotal role in the control
  of signal transduction by lipid mediator
  mediators, the ability of androgens to stimulate
  this enzyme in prostatic cells may provide
  opportunity for cross-talk between signaling
  pathways involving lipid mediators and androgens.

25
CSC Algorithm

• M(g,a) GO codes matched in article a for gene g
• O(g) GO codes annotated to the ortholog of g
• o size of O(g), p percentage (0.2)
• CSC(g,a)
  for every GO1 in M(g,a)
  count
  0
  for every GO2 in O(g)
  if((X2(GO1,GO2)gt3.84) (GO1 ne GO2))
  count
• if(count gt po)
  add GO1 to CSC(g,a)