Analysis of Maize Full-Length cDNAs -- a good example

Sequencing, Mapping, and Analysis of 27,455 Maize Full-Length cDNAs

Analysis

Contaminated FLcDNAs were found by comparing them against the maize, rice and Arabidopsis rRNA sequences with a BLAST e-value≤1e-50, which identified 26 rRNAs. An additional 110 FLcDNAs were identified that encoded proteins highly similar to bacteria (16 cDNAs), fungus (76 cDNAs) and vertebrate (18 cDNAs) and did not show similarity with plant proteins.

The ORFs were computed using the software GETORF in EMBOSS package [50] with parameters “–minsize 150, -find 1, -methionine, -noreverse”. TE and SSR analyses were performed using RepeatMasker (repeatmasker.org). For TE analysis, the Poaceae (grass family) TE database was downloaded from Genetic Information Research Institute (www.girinst.org) and the FLcDNAs that had masked sequence length of ≥100 bp were used for the TE insertion analysis. SSRs with length ≥20 bp and divergence ≤10% were selected for SSR location analysis. Putative transcription factors were analyzed using BLASTx with e-value≤1e-10 against rice and Arabidopsis transcription factor proteins downloaded from PlantTFDB (planttfdb.cbi.pku.edu.cn). Any maize cDNAs showing positive matches in both rice and Arabidopsis were assigned to TF families using the PlantTFDB nomenclature.

Plant homolog analysis was conducted using BLASTx (e-value≤1e-10) to compare rice, sorghum, Arabidopsis and poplar protein sequences downloaded from the following sites: 67,393 rice (MSU release 6.0; rice.plantbiology.msu.edu), 35,899 sorghum (www.phytozome.net/sorghum), 32,615 Arabidopsis (TAIR v8.0; www.arabidopsis.org) and 45,555 poplar (genome.jgi-psf.org). The maize FLcDNAs that did not have a homolog were compared with the plant UniProt database [29], where another 147 rice, sorghum, Arabidopsis or poplar homologs were identified and removed. Then the 1,475 putative unique maize FLcDNAs were mapped to GO annotated maize gene models with ≥95% ID and ≥90% alignment length using BLAT. GO over- and under- representation analysis were performed using Cytoscape [51] with BiNGO (Biological Networks Gene Ontology, [25]) plug-in and activating a hypergeometric distribution statistical test (p-value ≤0.05) with Benjamini and Hochberg false discovery rate (FDR) correction [52] relative to GO annotated maize gene models.

For annotation of all EST and FLcDNA assemblies, the unitrans were searched against the UniProt plants database (2009-06-17) using BLASTx with e-value≤1e-20. The GO [24]annotations were extracted from the UniProt file and gene association file (ftp.ebi.ac.uk/pub/databases/GO/goa/UNIP​ROT),which were mapped to plant GO Slim [32]. Some of the results were computed by custom Perl scripts, and the rest were obtained from the website, as follows: Table 6 was copied from the “Advanced Summary/Example Queries” page. The number of UniProt matches for the 27k were from the “UniTrans Search”, where “Non-maize UniProt Match” was set to ‘yes’; for the non-putative, the “Match Description” was set to “not putative”. Table 8, Table 9, and the top of Table 10 can all be verified from the PAVE query system.