pmic7840-sup-0010-test

advertisement
Materials and Methods:
Data: Mass spectral dataset for ME-AM1 was downloaded from PRIDE database [1]
and was part of the study by Schneider, K et. al.[2]. Briefly, Tandem mass spectrometry
data was generated from ME-AM1 grown on acetate and on methanol (Three biological
replicates for each substrate). Proteins from ME-AM1 cells were separated by 1D
SDS/PAGE and were followed after trypsin digestion on reverse phase HPLC coupled
to a LTQ-orbitrap high accuracy mass spectrometer [2]. Six spectral files with
accessions 17691-17696 were downloaded and converted to common search format
MGF(Mascot generic format) using PRIDEinspector tool [3]. In total 538,820 tandem
mass spectra were present in six files.
Genomic similarity/diversity estimation: dnadiff[4] utility from MUMmer suite was
used to compare genome sequences with default parameters. Genome identity
percentages for reference and query genes are the percentage of aligned bases
reported by dnadiff. While reporting average base pair identity between genomes, M to
M alignments were considered. The total non-redundant gene pool or Pan genome of
the Methylobacteriacea family was constructed by clustering all protein sequences from
individual methylobacterial genomes using CD-HIT[5] at 70% identity over 70% of query
sequence criterion. The cluster thus generated was processed using in-house Perl
scripts to generate a Phylogenetic Profile Matrix, where presence or absence of a
protein coding feature in any genome was represented by 1 or 0. Ultimately this matrix
was used to determine categorical existence of proteins. WebMGA[6] server was used
to annotate COG functional classes for a set of gene sequences.
Proteogenomic analysis: Raw mass spectra were searched against the reference
genome of ME-AM1 (NC_012808). Genome translation, peptide identification by
multiple search algorithms, FDR estimation and novel peptide identification were carried
out in an automated fashion by GenoSuite framework recently developed by our
group[7]. All four search algorithms, supported by GenoSuite, namely OMSSA[8],
X!Tandem[9], InsPecT[10] and MassWiz[11] were deployed for peptide identification.
Search parameters were 20 ppm precursor ion tolerance, 0.1 Da product ion tolerance,
trypsin as proteoase enzyme with one missed cleavage allowed, carbamydomethylation
of cysteine as fixed modification and oxidation of methionine was used as the variable
modification. Target database was composed of a six frame translated genome, proteins
from
the
other
four
replicons
and
common
contaminants
from
CRAP
(ftp://ftp.thegpm.org/fasta/cRAP) database. All 115 protein sequences from CRAP
database were included in the search database. All the sequences were reversed to
create the decoy database. A concatenated target-decoy search approach was used
and in total database had 118,664 protein entries. FDR (in %)for individual algorithmsis
calculated using following formula.
𝐹𝐷𝑅 =
𝐷
× 100
𝑇
D= count of decoy PSMs passing the threshold
T= count of target PSMs passing the threshold
Results from multiple algorithms were integrated using FDRscore[12] method. All these
steps are implemented in the GenoSuite pipeline. To infer proteins, identified peptides
were mapped back to the search database having 118,664 proteins. Only proteins with
≥2 peptides or single peptide hits with ≥5 significant PSMs were considered in the final
report.At this criterion protein identification FDR was calculated as follows
π‘ƒπ‘Ÿπ‘œπ‘‘π‘’π‘–π‘› 𝐹𝐷𝑅(%) =
𝐼𝑑𝑒𝑛𝑑𝑖𝑓𝑖𝑒𝑑 π‘‘π‘’π‘π‘œπ‘¦ π‘π‘Ÿπ‘œπ‘‘π‘’π‘–π‘› π‘π‘œπ‘’π‘›π‘‘
× 100
𝐼𝑑𝑒𝑛𝑑𝑖𝑓𝑖𝑒𝑑 π‘‘π‘Žπ‘Ÿπ‘”π‘’π‘‘ π‘π‘Ÿπ‘œπ‘‘π‘’π‘–π‘› π‘π‘œπ‘’π‘›π‘‘
Data visualisation: To analyze the conservation of novel proteins across closely
related genomes we used BRIG[13] tool with blastx option to search the genomes
against proteins. To visualize peptide identifications against the whole genome of MEAM1 a peptide co-ordinate file was created using in-house developed scripts and
Circos[14] was used for visualization of this data. To mark methylotrophy islands we
used methylotrophy gene information from Vuilleumier, S et. al. [15]. To visualize
genomic context of individual proteins identified from mass spectrometry data, we used
ORFmapper utility provided in GenoSuite. Peptide spectrum matches (PSMs) were
visualized for manual inspection using pLabel utility of pFind[16] studio with 0.5 Da
tolerance.
ORF prediction and function prediction: To impart additional confidence to novel
proteins abinitio predictions by GENEMARK, Prodigal, Glimmer and FgeneSB were
compared. This was done by comparing co-ordinates of gene predictions with that of
novel genesusing in-house scripts. The following steps describe how each of these
algorithms were employed for gene predictionο‚·
GeneMark – GeneMarkS (v4.6b) was used to create the training model and
prokaryotic GeneMark.hmm (v2.8) was employed to predict genes in ME-AM1
using the model obtained from GeneMarkS, while considering RBS for gene start
prediction.
ο‚·
Prodigal – version 2.60 was used with standard parameters to obtained gene
prediction.
ο‚·
Glimmer – The current version 3.02 of the software suite was obtained for
analysis. Using the “g3-from-scratch.csh” script of the glimmer suite, gene
prediction was performed on the nucleotide sequences for the genome of MEAM1. The program “g3-from-scratch.csh” runs a series of programs to
sequentially predict putative orfs to be used as a training set and to build model,
which is eventually used by the main program glimmer3 for the final stage of
gene prediction.
ο‚·
Fgenesb – the online implementation of fgenesb was used to predict genes in
ME-AM1 with BradyrhizobiumjaponicumUSDA110 as a close organism.
For functional prediction of proteins, PFAM (http://pfam.xfam.org/search#tabview=tab1)
and Goanna (http://www.agbase.msstate.edu/cgi-bin/tools/GOanna.cgi) web tools were
used with their default settings. For PFAM default settings include an e-value threshold
of 1.0 for pfam-A database and 0.001 for pfam-B search. Goanna parameters were
Database: UniProt, Expect: 10e-1, Matrix: BLOSUM62 and word size:3.
Normalized spectral abundance factor (NSAF): In-house scripts were used to
calculate NSAF for identified proteins in all six samples independently. For comparisons
between growth conditions, average NSAF was calculated for every protein.
To
calculate NSAF we followed the following formula proposed by Zybailov, B et. al. [17]
and considered only spectra which identified unique peptides at <1% FDR.
(𝑁𝑆𝐴𝐹)π‘˜ =
(𝑆𝑝𝐢/𝐿)π‘˜
𝑁
∑𝑖=1(𝑆𝑝𝐢/𝐿)𝑖
Where,
SpC= Total number of the qualified spectra assigned to protein
L= Length of the protein
N= Total number of proteins identified in the study
Reference List
1. Martens, L., Hermjakob, H., Jones, P., Adamski, M. etal., PRIDE: the proteomics
identifications database. Proteomics. 2005, 5, 3537-3545.
2. Schneider, K., Peyraud, R., Kiefer, P., Christen, P. etal., The ethylmalonyl-CoA
pathway is used in place of the glyoxylate cycle by Methylobacterium extorquens
AM1 during growth on acetate. J.Biol.Chem. 2012, 287, 757-766.
3. Wang, R., Fabregat, A., Rios, D., Ovelleiro, D. etal., PRIDE Inspector: a tool to
visualize and validate MS proteomics data. Nat.Biotechnol. 2012, 30, 135-137.
4. Kurtz, S., Phillippy, A., Delcher, A. L., Smoot, M. etal., Versatile and open software for
comparing large genomes. Genome Biol. 2004, 5, R12.
5. Li, W. and Godzik, A., Cd-hit: a fast program for clustering and comparing large sets of
protein or nucleotide sequences. Bioinformatics. 2006, 22, 1658-1659.
6. Wu, S., Zhu, Z., Fu, L., Niu, B., and Li, W., WebMGA: a customizable web server for
fast metagenomic sequence analysis. BMC.Genomics 2011, 12, 444.
7. Kumar, D., Yadav, A. K., Kadimi, P. K., Nagaraj, S. H. etal., Proteogenomic analysis of
Bradyrhizobium japonicum USDA110 using GenoSuite, an automated multialgorithmic pipeline. Mol.Cell Proteomics. 2013, 12, 3388-3397.
8. Geer, L. Y., Markey, S. P., Kowalak, J. A., Wagner, L. etal., Open mass spectrometry
search algorithm. J.Proteome.Res. 2004, 3, 958-964.
9. Craig, R. and Beavis, R. C., TANDEM: matching proteins with tandem mass spectra.
Bioinformatics. 2004, 20, 1466-1467.
10. Tanner, S., Shu, H., Frank, A., Wang, L. C. etal., InsPecT: identification of
posttranslationally modified peptides from tandem mass spectra. Anal.Chem.
2005, 77, 4626-4639.
11. Yadav, A. K., Kumar, D., and Dash, D., MassWiz: a novel scoring algorithm with
target-decoy based analysis pipeline for tandem mass spectrometry.
J.Proteome.Res. 2011, 10, 2154-2160.
12. Jones, A. R., Siepen, J. A., Hubbard, S. J., and Paton, N. W., Improving sensitivity in
proteome studies by analysis of false discovery rates for multiple search engines.
Proteomics. 2009, 9, 1220-1229.
13. Alikhan, N. F., Petty, N. K., Ben Zakour, N. L., and Beatson, S. A., BLAST Ring Image
Generator (BRIG): simple prokaryote genome comparisons. BMC.Genomics 2011,
12, 402.
14. Krzywinski, M., Schein, J., Birol, I., Connors, J. etal., Circos: an information aesthetic
for comparative genomics. Genome Res. 2009, 19, 1639-1645.
15. Vuilleumier, S., Chistoserdova, L., Lee, M. C., Bringel, F. etal., Methylobacterium
genome sequences: a reference blueprint to investigate microbial metabolism of
C1 compounds from natural and industrial sources. PLoS.One. 2009, 4, e5584.
16. Wang, L. H., Li, D. Q., Fu, Y., Wang, H. P. etal., pFind 2.0: a software package for
peptide and protein identification via tandem mass spectrometry. Rapid
Commun.Mass Spectrom. 2007, 21, 2985-2991.
17. Zybailov, B., Mosley, A. L., Sardiu, M. E., Coleman, M. K. etal., Statistical analysis of
membrane proteome expression changes in Saccharomyces cerevisiae.
J.Proteome.Res. 2006, 5, 2339-2347.
Download