Animal and Plant gene set reconstructions with EvidentialGene Comparisons to other popular and recent gene reconstructions In-progress, 2016/2017, D.G. Gilbert ------------------------------------------------- Tables summarizing alignment of test gene sets to reference species 1. Arabidopsis model plant 2. Corn (Zea mays) crop plant 3. White fly (Bemisia tabaci) crop plant pest 4. Water flea (Daphnia pulex) environmental sentinal species ------------------------------------------------- 1. Plant model Arabidopsis thal. gene reconstruction methods compared Gene assemblies of Illumina RNA-seq vs PacBio, 2017.feb, d.g.gilbert 1a. Gene assembly methods measured against unique coding sequences of reference Arabidopsis public gene set (2016, At16Ap, transcripts n=37806) Geneset nFound Found% AlignF% AlignT% At5EVm 36072 95.4 95.7 95.0 At2EVm 34294 90.7 94.0 92.3 Oases 34030 90.0 93.5 91.2 IDBAtr 33837 89.5 92.0 89.1 SOAPtr 33598 88.9 90.5 87.0 Trinity 33417 88.4 87.9 84.1 PacBio 21964 58.1 76.7 48.2 --------------------------- 1b. Arabidopsis gene sets measured against related species Orange and Cacao. Cacao Reference Orange Reference Geneset nGene nAlt Found% AlignF% nGene nAlt Found% AlignF% At16Ap 22334 131865 88.7 70.7 16795 22280 91.4 74.7 At17EVm5 22581 132507 89.1 70.3 16613 22087 90.6 74.2 At16Pacb 17593 95437 64.2 60.5 11578 15608 64.0 63.5 ---------------------------- --------------------------- 1c. Intron recovery for Arabidopsis gene sets (ni=125481 of RNA-seq mapped to chrs) Geneset GeneTr valExon Found% At16Ap 42211 110654 88.1 At17EVm5 141670 109859 87.5 At16Pacb 48848 70719 56.3 .. subset gene assemblies .. Oases 312817 101784 81.1 IDBAtr 248429 101357 80.7 SOAPtr 96533 99274 79.1 Trinity 198380 102208 81.4 ---------------------- Arabidopsis gene set versions At16Ap = public gene set of 2016 of Arabidopsis thal. from Araport.org Oases = velvet/oases assembly of Illumina RNA, IDBAtr = idba_tran asm of Ill. RNA, SOAPtr = SOAP-Trans asm of Ill. RNA, Trinity = Trinity asm of Ill. RNA, At16Pacb/PacBio = Pac-Bio "no-assembly" assembly (PacBio xxx method) of Pac-Bio RNA data At2EVm = Evigene classified gene set of Illumina RNA assemblies above (O,I,S,T) At17EVm5/At5EVm = Evigene reduction of assemblies of Illumina RNA-seq, improved over At2EVm with added RNA sets and gene assemblies At17EVm5 genes of 2017 are [soon to be] public at http://arthropods.eugenes.org/EvidentialGene/plants/arabidopsis/evigene2017_arabidopsis/ Method: BLASTn -query reference-unique.cds -db allgenesets.cds -evalue 1e-5 .. Statistics Found% = percent of reference transcripts found AlignF% = align to reference transcripts found. AlignT% = align to total reference transcripts. Intron Statistics and Methods, see below Reference genes: Orange, NCBI genomes/refseq/plant/Citrus_clementina/GCF_000493195.1_Citrus_clementina_v1.0 Cacao, Evigene update of Theobroma cacao from RNA-seq supplied by Mars. ============================================================= 2. Corn genes reconstructions 2a. Zea mays gene sets measured against reference proteins of Sorghum and Arabidopsis Sorghum proteins Arabidopsis proteins Geneset nFound Found% AlignF% AlignT% nFound Found% AlignF% AlignT% ZmEvig 28352 82.9 92.9 91.1 22727 79.5 86.7 85.6 ZmGram 28019 81.9 93.2 90.3 22661 79.2 86.2 84.8 ZmNCBI 27809 81.3 93.1 89.6 22640 79.2 86.2 84.7 ZmPacb 26680 78.0 89.3 82.4 22165 77.5 85.1 81.9 ZmJgi4 26557 77.6 88.4 81.2 21972 76.8 85.2 81.3 --------------------------- --------------------------- 2b. Restricted comparison to gene loci found in common over gene sets. Sorghum reference exons Arabidopsis reference exons Geneset nGene nAlt Found% AlignF% nGene nAlt Found% AlignF% ZmEvig 16905 17255 94.5 98.2 7472 7630 81.6 74.8 ZmGram 16905 17126 93.7 98.1 7472 7612 80.7 74.8 ZmPacb 16905 17194 90.2 97.2 7472 7617 80.4 74.7 ZmNCBI 16905 17046 91.6 98.0 7472 7538 78.5 74.6 ------------------------- --------------------------- 2c. Intron recovery for Zea mays gene sets (ni=198622 of RNA-seq mapped to chrs) Geneset GeneTr valExon Found% ZmEvig 171197 136609 68.7 ZmGram 143193 135333 68.1 ZmPacb 392946 135466 68.2 ZmJgi4 75241 136901 68.9 ZmNCBI na na na ---------------------- Corn gene sets ZmEvig = Evigene Zeamay5fEVm 2016 assembly of Illumina RNA-seq, public at http://arthropods.eugenes.org/EvidentialGene/plants/corn/evg5corn/ ZmGram = Ensembl/Gramene 2016.09 Zm000nnnn, ZmPacb = CSHL/Gramene PacBio gene assemblies of 2016 as SRA entries SRR3147024..054, ZmNCBI = NCBI 2014 refgen zeamay ZmJgi4 = JGI Rnnotator assembly set of Illumina RNA-Seq , 2014 Reference genes: Sorghum = Sorghum bicolor 2016, Sbicolor_313, from JGI/Phytozome.org, total primary isoforms = 34211 Arabidopsis = Arabidopsis thal. public gene set of 2015 from Araport.org, total primary isoforms = 28598 Method, 2a: BLASTp -query ref_proteins -db allgenesets_proteins -evalue 1e-5 .. Statistics, 2a (more at end) Found% = percent of reference transcripts found AlignF% = align to reference transcripts found. AlignT% = align to total reference transcripts. Method, 2b: BLASTn -query reference-unique.cds -db allgenesets.cds -evalue 1e-5 .. Statistics, 2b nAlt = number of reference alternate transcripts found, for nGene reference loci, restricted to reference loci common to all gene sets. Found% = percent of reference exons found, of total ref unique exons AlignF% = align to ref exons found, as percent of ref exon sizes. Method and Statistics, 2c : see below ============================================================= 3. White fly Bemisia tabaci (cotton/crop plant pest) Gene sets of White fly measured against reference proteins Reference Pea aphid Fruit fly Conserved F.Fly Geneset Found% AlnF% AlnT% Found% AlnF% AlnT% Found% AlnF% AlnT% BtEvig 81.2 90.0 88.0 74.1 81.8 74.9 98.7 85.4 80.6 BtNCBI 79.7 85.5 82.3 73.4 78.7 71.6 98.5 83.0 78.1 BtMakr 77.4 80.2 73.8 72.1 74.9 66.0 97.1 79.2 72.9 BtTrin 73.5 71.0 59.2 68.0 67.4 53.2 95.5 74.4 64.1 ---------------- ----------------- ----------------- 3b. Intron recovery for Bemisia tabaci gene sets (ni=134153 of RNA-seq mapped to chrs) Geneset GeneTr valExon Found% BtEvig 63368 91946 68.5 BtNCBI 20506 93157 69.4 BtMakr 13825 77455 57.7 BtTrin 20534 67879 50.5 ----------------------- Bemisia tabaci gene sets compared BtEvig = Evigene gene assembly, 2016 update (vers 3), available [soon] at http://arthropods.eugenes.org/EvidentialGene/arthropods/whitefly/whitefly3evigene/ BtNCBI = NCBI RefSeq gene models, 2016 BtMakr = Whitefly genome project genes modeled with MAKER, 2016, whiteflygenomics.org BtTrin = TSA.GBII gene assembly 2015, Trinity of Illumina Reference genes: Pea aphid Acyr. pisum, NCBI RefSeq 2016, total primary isoforms n=18601 Fruit fly, NCBI RefSeq 2015, total primary isoforms n=13828 Conserved Fruit fly, NCBI RefSeq 1-copy genes identified by BUSCO, total n=3055 Method: BLASTp -query ref_proteins -db allgenesets_proteins -evalue 1e-5 .. Statistics Found% = percent of reference proteins w/ signif. alignment AlnF,AlignF% = average percent align to found reference protein AlnT,AlignT% = percent align to all reference proteins ============================================================= 4. Water flea Daphnia pulex gene sets compared Reference Daphnia magna Fruit fly Conserved F.Fly DpEvig7 DpMaker7 DpEvig7 DpMaker7 DpEvig7 DpMaker7 Found% 72.0 58.9 67.9 64.3 94.9 90.3 AlignF% 88.8 85.6 80.7 79.1 83.7 81.9 AlignT% 88.6 69.9 80.3 74.5 83.5 77.7 Tiny% 0.2 23.8 0.8 6.5 na Best% 75.4 3.1 65.4 5.2 na equal 21.3 equal 29.2 ------------------------ --------------- ---------------- 4b. Intron recovery for Daphnia pulex gene sets (ni=189710 of RNA-seq mapped to chrs) Geneset GeneTr valExon Found% DpEvig7 115051 126417 66.6 DpMaker7 18296 88747 46.7 ---------------------- Daphnia pulex gene sets DpEvig7 Evigene genes of 2017 from http://arthropods.eugenes.org/EvidentialGene/daphnia/daphnia_pulex/daphnia_pulex_genes2017/ DpMaker7 genes of 2017 from report of doi:10.1534/g3.116.038638 Reference genes: Daphnia magna water flea, Evigene 2015 set, primary isoforms n=29127 Drosophila melanogaster fruit fly, NCBI RefSeq 2015, total primary isoforms n=13828 Conserved Fruit fly, NCBI RefSeq 1-copy genes identified by BUSCO, total n=3055 Methods: BLASTP -query reference.aa -db dapplx_twogenesets.aa -evalue 1e-5 Statistics: Found = percent of reference genes with signif. align in target gene set AlignF = average % alignment to found reference genes (align-aa/ref-aa) Tiny = target genes with size < 50% of reference length, of found genes Best = which target set has longest alignment per ref gene, of found genes ============================================================= Common methods and statistics BLASTp|BLASTn -query reference.sequences -db allgenesets.sequences -evalue 1e-5 Tabulate largest alignement of each target set to each reference sequence (targets matching 2+ refs allowed), count and sum significant alignments per gene set, including Found% = found alignments, 100 * n-aligned/n-ref; AlignF% = sum(100*align-bases/ref-bases)/nFound, average of per-ref percent align, for found alignments; AlignT% = 100*sum(aligned-bases)/sum(ref-bases) Tiny% = 100 * count(n-target-size < 50% of ref-size) / nFound, for aligned target Best% = pairwise measure for 2 gene sets only, counts of target-A <=> target-B alignment per reference bestA = sum(alignA > minDiff + alignB)/nFound(A or B), minDiff = 9 or other bestB = sum(alignB > minDiff + alignA)/nFound(A or B) equal = sum(abs(alignA - alignB) <= minDiff)/nFound(A or B) Two independent quality measures of gene sets are (a) reference genes found or reconstructed in some fashion, (b) completeness of reconstruction of reference genes (full alignment with target length =~ reference length). AlignF is this (b) independent measure from (a) Found genes, while AlignT is roughly the union of these. There is a subtle bias in AlignF: those target gene sets with fewer found genes tend to find the longer, better aligned reference genes. AlignF can be larger where fewer are found, even when alignments are identical between target sets for found genes. Pairwise comparisons of 2 gene sets, i.e. the Best metric, is not biased by the different reference subsets found between gene sets. Additional measures of target gene sizes vs reference size, independent of alignment size, are useful; In practice, small outliers Tiny%, indicate fragmented gene prevelance and can help distinguish gene set qualities. Large outliers, indicative of some kinds of joined-gene artifacts, don't usually discriminate gene sets, but may depending on how a gene set was constructed. Intron Methods and Statistics map RNA-seq (Illumina) to chromosome assembly with GSNAP, extract splice-mapped reads and their intron locations, tabulate gene-exon x rna-intron matches. Statistics GeneTr = gene transcripts total in gene set valExon = gene exons w/ validated intron InFound% = percent of all valid introns recovered b/n gene exons =============================================================