1. Searching for Transcription Start Sites in Drosophila
Wilson Leung
08/2017

2. Outline Transcription start sites (TSS) annotation goals
Promoter architecture in D. melanogaster
New D. melanogaster TSS datasets
Find the initial transcribed exon
Annotate putative transcription start sites
Search for core promoter motifs

3. Muller F element, heterochromatic, and euchromatic genes show similar expression levels
F element genes show lower expression levels in S2 (late embryos) cells but no difference in BG3 (neuronal) cells
Position affects gene expression
See Dr. Elgin’s presentation on the private wiki for a more detailed explanation of the scientific goals.
Riddle NC, et al. PLoS Genet Sep;8(9):e

4. TSS of F element genes show lower levels of H3K9me3 and HP1a
POF
Su(var)3-9
PolII
H3K36me3
POF
PolII
HP1a
Su(var)3-9
H3K9me3 over the gene body
H3K36me3 – associated with 3’ end of transcript in euchromatic regions
Riddle NC, et al. PLoS Genet Sep;8(9):e

5. Three strategies for motif finding
Multiple genes in a single species
Genes with common expression pattern
Sequences associated with ChIP-Seq peaks
Single gene in multiple species
Phylogenetic footprinting
Multiple genes in multiple species
Compare multiple sequence alignment profiles of multiple genes (Magma)

6. Motif finding using multiple genes within a single species1 2 3 4 5 6 7 8 9 10 A 12 13 C 22 23 64 70 54 33 G 11 30 14 T 36 31 56 40 19
Bits
0.0
1.0
2.0 5 10
Trl: FlyReg_DNaseI
Zero or one instances per sequence
Sequences surrounding TSS
Predicted motif instances

7. Motif finding using single gene in multiple species
Genes
PhyloP
phastCons
Conserved Elements
Multiple Sequence Alignment
D. mel: chr4
EvoprinterHD

8. Motif finding using multiple genes in multiple species (PhyloNet)
1. Identify conserved regions (profiles) in whole genome multiple sequence alignments
2. Identify multiple genes in the genome with similar alignment profiles
Create phylogenetic profiles for each promoter
Modify Karlin-Altschul statistics to calculate E-values (compare homologous sequences)
Cluster HSPs into profile alignments and generate final motifs
Create a new continuous profile space (15 subprofile spaces – tetrahedron where the four vertices represent the nucleotides)
Promoter sequences
Conserved motifs
Based on Figure 1 from Wang T and Stormo GD. PNAS 2005 Nov 29;102(48):

9. Magma: Multiple Aligner of Genomic Multiple Alignments
Key features of Magma:
Runs ~70x faster than PhyloNet
Analyze multiple sequence alignments with gaps
Use set-covering approach to minimize redundancy in discovered motifs
Comparison using average log likelihood scores (ALLR) - extended to allow for gaps
Computationally tractable to analyze conserved motifs in multiple eukaryotic genomes
Ihuegbu NE, Stormo GD, Buhler J. J Comput Biol Feb;19(2):

10. Goals for the transcription start sites (TSS) annotations
Research goal:
Identify motifs that enable Muller F element genes to function within a heterochromatic environment
Annotation goals:
Define search regions enriched in regulatory motifs
Define precise location of TSS if possible
Define search regions where TSS could be found
Document the evidence used to support the TSS annotations
Detailed documentations allow us to prioritize the list of TSS candidates

11. Estimated evolutionary distances with respect to D. melanogaster
D. simulans
Species
Substitutions per neutral site
D. ficusphila
0.80
D. eugracilis
0.76
D. biarmipes
0.70
D. takahashii
0.65
D. elegans
0.72
D. rhopaloa
0.66
D. kikkawai
0.89
D. bipectinata
0.99
D. sechellia
D. yakuba
D. erecta
D. ficusphila
D. eugracilis
D. biarmipes
D. takahashii
D. elegans
D. rhopaloa
D. kikkawai
D. bipectinata
D. eugracilis – training project
D. ficusphila – higher substitutions per neutral sites than D. biarmipes and D. elegans
Purple = priority 1 species in modENCODE white paper
D. melanogaster subgroup = 0.4ss
D. ananassae = 1.3ss
D. ananassae
D. pseudoobscura
D. persimilis
D. willistoni
Data from Table 1 of the modENCODE comparative genomics white paper
D. mojavensis
D. virilis
D. grimshawi
GEP annotation projects
Species sequenced by modENCODE

12. Challenges with TSS annotations
Fewer constraints on untranslated regions (UTRs)
UTRs evolve more quickly than coding regions
Open reading frames, compatible phases of donor and acceptor sites do not apply to UTRs
Low percent identity (~50-70%) between D. biarmipes contigs and D. melanogaster UTRs
Most gene finders do not predict UTRs
Lack of experimental data
Cannot use RNA-Seq data to precisely define the TSS
Similar levels of sequence similarity between D. elegans and D. melanogaster

13. TSS annotation workflow
Identify the ortholog
Note the gene structure in D. melanogaster
Annotate the coding exons
Classify the type of core promoter in D. melanogaster
Annotate the initial transcribed exon
Identify any core promoter motifs in region
Define TSS positions or TSS search regions
Annotation is based on parsimony with D. melanogaster

14. RNA Polymerase II core promoter
Initiator motif (Inr) contains the TSS
TFIID binds to the TATA box and Inr to initiate the assembly of the pre- initiation complex (PIC)
polypyrimidine initiator (TCT) motif associated with ribosomal genes
Core promoter 200bp surrounding the TSS
K (keto) = G or T, W (weak) = A or T
Mammalian genes have CpG islands but Drosophila promoters do not
Juven-Gershon T and Kadonaga JT. Regulation of gene expression via the core promoter and the basal transcriptional machinery. Dev Biol Mar 15;339(2):225-9.

15. Peaked versus broad promoters
Peaked promoter
(Single strong TSS)
Broad promoter
(Multiple weak TSS) bp
Peaked promoter in D. melanogaster is more informative: expect peak promoter in the target species
focused (peaked) versus dispersed (broad) promoter
Motifs associated with peaked promoters have more well-defined positions relative to +1
Kadonaga JT. Perspectives on the RNA polymerase II core promoter. Wiley Interdiscip Rev Dev Biol Jan-Feb;1(1):40-51.

16. RNA-Seq biases introduced by library construction
cDNA fragmentation
Strong bias at the 3’ end
RNA fragmentation
More uniform coverage
Miss the 5’ and 3’ ends of the transcript
RNA-Seq Read Count
This chart uses yeast (single ORF) as an example
It does not account for biases in read mapping
RNA fragmentation (e.g., hydrolysis, nebulization) missed the transcript ends
cDNA fragmentation (e.g., sonication, DNaseI) has strong bias at the 3’ end
Transcripts show more uniform coverage using RNA instead of cDNA fragmentation 5’ 3’
Gene Span
Wang Z, et al. RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet Jan;10(1):57-63.

17. Techniques for finding TSS
Identify the 5’ cap at the beginning of the mRNA
Cap Analysis of Gene Expression (CAGE)
RNA Ligase Mediated Rapid Amplification of cDNA Ends (RLM-RACE)
Cap-trapped Expressed Sequence Tags (5’ ESTs)
More information on these techniques:
Takahashi H, et al. CAGE (cap analysis of gene expression): a protocol for the detection of promoter and transcriptional networks. Methods Mol Biol :
Sandelin A, et al. Mammalian RNA polymerase II core promoters: insights from genome-wide studies. Nat Rev Genet Jun;8(6):
CAGE randomly sample 5’ end – correspond to expression level
RACE target 5’ ends of specific transcripts (8727 transcripts) – capture lowly expressed transcripts
5’ end of mRNA has a guanine connected to the rest of the mRNA via a 5’ to 5’ triphosphate link.
This guanosine is methylated at position 7 (7-methylguanylate cap) and helps increase the stability of the mRNA.

18. Promoter architecture in Drosophila
Classify core promoter based on the Shape Index (SI)
Determined by the distribution of CAGE and 5’ RLM-RACE reads
Shape index is a continuum
Most promoters in D. melanogaster contain multiple TSS
Median width = 162 bp
~70% of vertebrate genes have broad promoters
median width 162bp => approximately the length of DNA for a single nucleosome
Most of the promoters in Drosophila are broad promoters
Motif associated with Paused polymerase are found in peaked promoters
Hoskins RA, et al. Genome-wide analysis of promoter architecture in Drosophila melanogaster. Genome Res Feb;21(2):

19. Genes with peaked promoters show stronger spatial and tissue specificity
46% of genes with broad promoters are expressed in all stages of embryonic development
19% of genes with peaked promoters are expressed in all stages
Peaked promoter – have more precise control
Hoskins RA, et al. Genome-wide analysis of promoter architecture in Drosophila melanogaster. Genome Res Feb;21(2):

20. Peaked and broad promoters are enriched in different core promoter motifs
Rach EA, et al. Motif composition, conservation and condition-specificity of single and alternative transcription start sites in the Drosophila genome. Genome Biol. 2009;10(7):R73.

21. Resources for classifying the type of core promoter in D. melanogaster
Only a subset of the modENCODE data are available through FlyBase
D. melanogaster GEP UCSC Genome Browser [Aug (BDGP Release 6) assembly]
FlyBase gene annotations (release 6.16)
modENCODE TSS (Celniker) annotations
DNase I hypersensitive sites (DHS)
CAGE and RAMPAGE TSS datasets
9-state and 16-state chromatin models
Transcription factor binding site (TFBS) HOT spots
16-state hiHMM models: late embryos and third instar larvae
9-state chromHMM models: S2 and BG3 cells
DHS data available for different embryonic stages and cell lines

22. 9-state chromatin model
Kharchenko PV, et al. Comprehensive analysis of the chromatin landscape in Drosophila melanogaster. Nature Mar 24;471(7339):480-5.

23. DNaseI Hypersensitive Sites (DHS) correspond to accessible regions
Aasland R and Stewart AF. Analysis of DNaseI hypersensitive sites in chromatin by cleavage in permeabilized cells. Methods Mol Biol. 1999;119:
Ho JW, et al. Comparative analysis of metazoan chromatin organization. Nature Aug 28;512(7515):

24. modENCODE TSS annotations
Two sets of modENCODE TSS predictions
TSS (Celniker)
Most recent dataset produced by modENCODE
Available on the GEP UCSC Genome Browser
TSS (Embryonic)
Older dataset available from FlyBase GBrowse
Use TSS (Celniker) dataset as the primary evidence
BDGP working with FlyBase to update the TSS annotations
Updated RNA-Seq datasets available since release 6.03
Hoskins RA, et al. Genome-wide analysis of promoter architecture in Drosophila melanogaster. Genome Res Feb;21(2):182-92

25. Classify the D. melanogaster core promoter based on (TSS) Celniker annotations and DHS positions
TSS classification
# Annotated TSS
# DHS positions
Peaked 1
Intermediate
≤ 1
> 1
Broad
Insufficient evidence
Updated definition for peaked and intermediate TSS to cover all cases
Classification is based on each unique TSS
Different unique TSS for the same gene could have different classifications
Consider DHS positions within a 300bp window surrounding the start of the D. melanogaster transcript

26. DEMO: Classify the core promoter of D. melanogaster Rad23

27. Additional DHS data from different stages of embryonic development
DHS data produced by the BDTNP project
Evidence tracks:
Detected DHS Positions (Embryos)
DHS Read Density (Embryos)
BG3 9-state
S2 9-state
chr4
CG2316-RB
CG2316-RD
CG2316-RC
CG2316-RA
CG2316-RG
CG2316-RH
BG3 DHS
S2 DHS
Kc DHS
BDTNP = Berkeley Drosophila Transcription Network Project
Use DHS data from cell lines as primary evidence
BG3 = CNS 3rd instar larvae, S2 = late embryos, Kc167 = dorsal closure stage
stage 14 = min; Dorsal closure of midgut and epidermis
Determine expression pattern of the gene in D. melanogaster through the “High-Throughput Expression Data” section
Stage 5
Stage 9
Stage 10
Stage 11
Stage 14
TSS (Celniker)
Thomas S, et al. Dynamic reprogramming of chromatin accessibility during Drosophila embryo development. Genome Biol. 2011;12(5):R43.

28. Additional TSS data available in FlyBase release 6.11
MachiBase
Batut P, Dobin A, Plessy C, Carninci P, Gingeras TR. High-fidelity promoter profiling reveals widespread alternative promoter usage and transposon-driven developmental gene expression. Genome Res Jan;23(1):

29. Benefits of RAMPAGE
RAMPAGE = RNA Annotation and Mapping of Promoters for Analysis of Gene Expression
CAGE only allows sequencing of short sequence tags (~27 bp) near the 5’ cap
Ambiguous read mapping to large parts of the genome
RAMPAGE produces long paired-end reads instead of short sequence tags
Developed novel algorithm to identify TSS clusters
Used paired-end information during peak calling
Used Cufflinks to produce partial transcript models
CAGE = Cap analysis gene expression
RAMPAGE = combine template switching and cap trapping
template switching = add adapter to end of 5’ complete first-strand cDNAs during reverse transcription
cap trapping = biotinylation of the 7-methylguanosine cap of Pol II transcripts and pull down of 5’-complete cDNAs
MachiBase – based on modified version of SAGE (Serial Analysis of Gene Expression)
Less specific than CAGE or RAMPAGE
Batut P, Gingeras TR. RAMPAGE: promoter activity profiling by paired-end sequencing of 5'-complete cDNAs. Curr Protoc Mol Biol Nov 11;104:Unit 25B.11.

30. RAMPAGE results on the GEP UCSC Genome Browser
Lifted RAMPAGE results from release 5 to release 6
Results from 36 developmental stages
Combined TSS peak call from all samples
Available under the “Expression and Regulation” section

31. Standardize analysis of MachiBase and modENCODE CAGE data using CAGEr
Bioconductor package developed by RIKEN
Map datasets against release 6 assembly
37 modENCODE CAGE samples; 7 MachiBase samples
Define TSS and promoters for each sample
Define consensus promoters across all samples
MachiBase = 5’ SAGE data
Haberle V, et al. CAGEr: precise TSS data retrieval and high-resolution promoterome mining for integrative analyses. Nucleic Acids Res Apr 30;43(8):e51.

32. TSS classifications based on CAGEr
Peaked
FlyBase Genes
modENCODE CAGE Peaks
modENCODE CAGE (Plus)
Intermediate
FlyBase Genes
modENCODE CAGE Peaks
modENCODE CAGE (Plus)
Broad
modENCODE CAGE Peaks
modENCODE CAGE (Minus)
FlyBase Genes

33. Changes in the dominant TSS of Rad23 across different developmental stages
CAGE Tag Clusters
Tag cluster interval ~70bp, 80% of signal within 50bp
thin box = interquantile width: 80% of CAGE signal
BG3 cells L3 digestive system, also show different dominant TSS
Stages of Development
Adult females

34. Evidence for TSS annotations (in general order of importance)
Experimental data
RNA-Seq
RNA Pol II ChIP-Seq
Conservation
Type of TSS (peaked/intermediate/broad) in D. melanogaster
Sequence similarity to initial exon in D. melanogaster
Sequence similarity to other Drosophila species (Multiz)
Core promoter motifs
Inr, TATA box, etc.
Use D. biarmipes PolII data to help define the TSS search regions in the other species

35. Determine the gene structure in D. melanogaster
UTR
CDS
FlyBase: GBrowse
Gene Record Finder: Transcript Details

36. Identify the initial transcribed exon using NCBI blastn
Retrieve the sequences of the initial exons from the Transcript Details tab of the Gene Record Finder
Use placement of the flanking exons to reduce the size of the search region if possible
Increase sensitivity of nucleotide searches
Change Program Selection to blastn
Change Word size to 7
Change Match/Mismatch Scores to +1, -1
Change Gap Costs to Existence: 2, Extension: 1

37. Extrapolate TSS position based on blastn alignment of the initial transcribed exon
blastn: D. mel: Rad23:1 (Query) vs. contig19 (Sbjct)
Query start: 6
Extrapolate TSS position:
28,941-5 = 28,936
Assume the length of the initial transcribed exon is conserved between D. melanogaster and D. biarmipes

38. Core promoter motifs can affect gene expression levels
SCP1:
SCP1 = Super Core Promoter 1
mTATA = variant of SCP1 with mutations in TATA
Juven-Gershon T, et al. Rational design of a super core promoter that enhances gene expression. Nat Methods Nov;3(11):

39. Use core promoter motifs to support TSS annotations
Some sequence motifs are enriched in the region (~300 bp) surrounding the TSS
Some motifs (e.g., Inr, TATA) are well-characterized
Other motifs are identified based on computational analysis
Presence of core promoter motifs can be used to support the TSS annotations
Inr motif (TCAKTY) overlaps with the TSS (-2 to +4)
Absence of core promoter motifs is a negative result
Most D. melanogaster TSS do not contain the Inr motif
TCA for broad promoters

40. Use UCSC Genome Browser Short Match to find Drosophila core promoter motifs
Ohler U, et al. Computational analysis of core promoters in the Drosophila genome. Genome Biol. 2002; 3(12):RESEARCH0087.
TATA box
Initiator (Inr)
K = keto (G or T), Y = Pyrimidine (C or T)
Consider removing BREu and BREd from list because of large number of false positives
Available under “Projects”  “Annotation Resources”  “Core Promoter Motifs” on the GEP web site:

41. Core Promoter Motifs tracks
Show core promoter motif matches for each contig
Separated by strand
Visualize matches to different core promoter motifs
Use UCSC Table Browser (or other means) to export the list of motif matches within the search region
Consider ways to generate locations of core promoter motifs in D. melanogaster

42. DEMO: Use the Inr motif to support the TSS position of Rad23

43. RNA PolII ChIP-Seq tracks (available for D. biarmipes, D
RNA PolII ChIP-Seq tracks (available for D. biarmipes, D. elegans, and D. ficusphila)
Show regions that are enriched in RNA Polymerase II compared to input DNA
Gene Models
RNA PolII Peaks
RNA PolII Enrichment
RNA-Seq
Perform blastn search of D. biarmipes region enriched in RNA PolII against the target species

44. Narrow TSS search region
Using RNA-Seq and RNA PolII ChIP-Seq data to define the TSS search region
D. mel Transcripts
RNA-Seq
Could define narrow and wide search regions to encompass regions supported by weak TSS evidence
RNA PolII Peaks
RNA PolII Enrichment
Narrow TSS search region

45. TSS annotation for Rad23 TSS position: 28,936
Conservation with D. melanogaster
blastn search of initial exon
“D. mel Transcripts” track
Location of the Inr motif
TSS search region: 28,716-28,936
Enrichment of RNA PolII upstream of the TSS position
RNA-Seq read coverage upstream of the TSS position
Search region defined by the extent of the RNA PolII peak

46. TSS annotation resources
Walkthroughs:
Annotation of Transcription Start Sites in Drosophila
Sample TSS report for onecut
Reference:
TSS Annotation Workflow
GEP Annotation Report:
Classify the type of core promoter
Evidence that supports or refutes the TSS annotation
Distribution of core promoter motifs
Additional curriculum on motif finding also available under the “Beyond annotation” section
Added new parameters page to the TSS Annotation Workflow

47. Additional TSS annotation resources
The D. melanogaster gene annotations are the primary source of evidence
Resources that could be useful if the D. melanogaster evidence is ambiguous
Whole genome alignments of multiple Drosophila species
PhastCons and PhyloP conservation scores
Genome browsers for nine Drosophila species
RNA Pol II ChIP-Seq (D. biarmipes, D. elegans, and D. ficusphila)
RNA-Seq coverage, TopHat junctions, assembled transcripts
Augustus and N-SCAN gene predictions
Cross-species alignments of Gnomon gene predictions
onecut example: supported by both RNA-Seq and Multiz

48. TSS annotation summary
Most of the D. melanogaster core promoters have multiple TSS
Classify the type of promoter (peaked/intermediate/broad) based on the transcriptome evidence from D. melanogaster
Define search regions that might contain TSS
Use multiple lines of evidence to infer the TSS region
Identify initial exon
RNA-Seq coverage
blastn (change search parameters)
Distribution of core promoter motifs (e.g., Inr)
RNA PolII ChIP-Seq peaks
Maintain conservation compared to D. melanogaster

49. Questions?

51. Structure of a typical mRNA
Pesole G. et al. Untranslated regions of mRNAs. Genome Biology. 2002: 3(3) reviews reviews

52. Expression Levels (rlog)
D. ananassae and D. melanogaster F element genes show similar range of expression levels
Adult Females
Adult Males
F element
D. mel: F (modENCODE) 4
D. ana: F (modENCODE) 4L 4R
Chen Z-X, et al. Genome Res. 24:
D. ananassae Adult Females
CAI (Codon Bias)
Expression Levels (rlog)
LOESS Regression Line

53. Phylogenetic tree based on the analysis of 13 Type IIB restriction endonucleases
D. simulans
Simulate restriction digests of 21 genomes
DNA fragments range from bp in size
Calculate distance between two genomes based on number of shared fragments
D. sechellia
D. melanogaster
D. yakuba
D. santomea
D. erecta
D. eugracilis
D. biarmipes
D. takahashii
D. elegans
D. rhopaloa
D. ficusphila
Recognition sequence of Type IIB endonucleases are 5-7bp long and cut dsDNA
D. kikkawai
D. ananassae
D. bipectinata
D. persimilis
D. pseudoobscura
Seetharam AS and Stuart GW. Whole genome phylogeny for 21 Drosophila species using predicted 2b-RAD fragments. PeerJ Dec 23;1:e226.
D. willistoni
D. virilis
D. mojavensis
D. grimshawi

54. RAMPAGE protocol
Batut P, et al. High-fidelity promoter profiling reveals widespread alternative promoter usage and transposon-driven developmental gene expression.
( supplementary figure 1.
Batut P, Gingeras TR. RAMPAGE: promoter activity profiling by paired-end sequencing of 5'-complete cDNAs. Curr Protoc Mol Biol Nov 11;104:Unit 25B.11.
Ribosome-depleted RNA is reverse-transcribed with random primers bearing an Illumina adaptor sequence overhang. Under the conditions used, the reverse transcriptase will often add a few non-templated Cs when it reaches the 5′ end of the template, especially if the template is capped. A template-switching oligo (TSO), which has three riboguanosines at its 3′ end, can hybridize to the terminal Cs, prompting the enzyme to switch templates and add the TSO sequence to the end of the newly synthesized cDNA. Since the TSO bears the other Illumina adaptor sequence, resulting 5′-complete cDNAs are amplifiable, whereas non-5′-complete molecules are not. The next steps implement the cap-trapping strategy, in which riboses with free 2′- and 3′-hydroxyl groups are oxidized and biotinylated, and single-stranded portions of RNA are digested by RNase I. This leaves biotin groups at only the 5′ ends of capped transcripts hybridized to 5′-complete cDNAs, which can then be recovered on streptavidin-coated beads. After PCR amplification and size selection, the cDNAs selected by these two orthogonal strategies can be directly sequenced on Illumina platforms.

55. “FlyBase: GBrowse Tracks” page on the FlyBase Wiki
Signals in the FlyBase RAMPAGE and MachiBase TSS tracks are off by one base
“FlyBase: GBrowse Tracks” page on the FlyBase Wiki

56. DEMO
blastn search of the initial transcribed exon of Rad23 against D. biarmipes contig19

57. Optimize alignment parameters based on expected levels of conservation
Derive alignment scores using information theory
Relative entropy of target and background frequencies
Match +2, Mismatch -3 optimized for 90% identity
Match +1, Mismatch -1 optimized for 75% identity
Less information available per aligned position
WU-BLAST: +5/-4, +5,-11
information = decrease in uncertainty
Convey more information from larger vocabulary and surprising answer (inversely proportional to its probability)
PAM = Point Accepted Mutations
Need aligned base to have some minimum information content to get significant alignment
Shannon Entropy = unpredictability of random variable
= subtract Kullback-Leibler divergence (target vs. uniform distribution) from total entropy required to encode a message
States DJ, et al. Improved Sensitivity of Nucleic Acid Database Searches Using Application-Specific Scoring Matrices. Methods :66-70.

58. Use RNA PolII tracks on the D
Use RNA PolII tracks on the D. biarmipes genome browser to identify putative TSS
April 2013 (BCM-HGSC/Dbia_2.0) assembly
Search for orthologous regions in D. elegans
Use more stringent parameters than the GEP annotation projects (problem with overlapping projects).

59. Gnomon predictions for eight Drosophila species
Based on RNA-Seq data from either the same or closely-related species
D. simulans, D. yakuba, D. erecta, D. ananassae, D. pseudoobscura, D. willistoni, D. virilis, and D. mojavensis
Predictions include untranslated regions and multiple isoforms
Records not yet available through the NCBI RefSeq database
Access these annotations through the FlyBase BLAST service

60. Conservation tracks on the D. melanogaster GEP UCSC Genome Browser
Whole genome alignments of multiple Drosophila species
Drosophila Chain/Net composite track
Generate multiple sequence (Multiz) alignments from these pairwise alignments
Identify conserved regions from Multiz alignments
PhastCons: identify conserved elements
PhyloP: measure level of selection at each nucleotide
Multiz alignment of 27 insect species available on the official UCSC Genome Browser
Aug (BDGP Release 6 + ISO1 MT/dm6) assembly

61. Use the conservation tracks to identify regions under selection
PhyloP scores:
Under negative selection
Fast-evolving

62. Examine the Multiz alignments to identify the orthologous TSS regions

63. Use RNA-Seq data to predict untranslated regions and putative TSS
TSS predictions available for 9 Drosophila species
N-SCAN+PASA-EST, Augustus, TransDecoder
D. mel Proteins
N-SCAN
Augustus
TransDecoder
RNA-Seq