data analysis developed in MatLab at Stanford University. Peak
selection was handled automatically within Byonic.
2. MS/MS data were searched against a UniProtKB FASTA database containing 16,972 reviewed Mus musculus entries (various
dates). Propionamidation (+71.037114 @ C) was set as a fixed
modification, Deamination (+0.984016 @ N) and Acetylation
(+42.010565 @ K) were set as common1 modifications, Oxidation (+15.994915 @ M) was set as common2 modification,
and Acetylation (+42.010565 @ Protein N-term) and Methylation (+14.01565 @ K, R) were set as rare1 modifications.
Byonic was set to allow a maximum of two common modifications and one rare modification and to allow a maximum of two
missed cleavages. MS/MS spectra were matched with a tolerance of 12 ppm on precursor mass and 0.4 Da on
fragment mass.
3. Common contaminants were filtered automatically by Byonic
and include TRYP_PIG, ALBU_BOVIN, ALBU_HUMAN,
CASB_BOVIN, CASK_BOVIN, CAS1_BOVIN, CO3_HUMAN, HBA_HUMAN, HBB_HUMAN, K1M1_SHEEP,
K2C1_HUMAN,
K22E_HUMAN,
K1C10_HUMAN,
K1C15_SHEEP,
K1C9_HUMAN,
KRHB1_HUMAN,
KRHB3_HUMAN, KRHB5_HUMAN, KRHB6_HUMAN,
and TRFE_HUMAN.
4. Using Byonic, the proteome was searched with a reverse-decoy
strategy and all data were filtered and presented at a 1% false
discovery rate. Byonic calculates a Byonic score that is an
indicator of the correctness of our peptide-spectrum matches
(PSM). Byonic scores reflect the absolute quality of the PSM.
Byonic scores range from 0 to 1000, with 300 being a good
score, 400 a very good score, and scores over 500 reflecting
near-perfect matches. For our data, all filtered protein identification hits have an FDR rate lower or equal to 2.5%, a Byonic
score greater than 250, and a log probability greater than 3.
5. To further discriminate nonsignificant to statistically significant
proteins, we also recommend running a statistical power
analysis [32].
6. Once the statistically significant proteins are determined, a
spectral counting method known as normalized spectral abundance factor (NSAF) [33] can be used to determine the relative
abundance of proteins in the samples [27, 33, 34], and to verify
the reproducibility, between replicates, of the quantitative data
through a correlation analysis [34].
7. Once the statistically significant proteins are determined, along
with their relative abundance, a number of analyses can be
performed to look at functional and/or localization differences
40
Ana Gordon and Karine Gousset
selection was handled automatically within Byonic.
2. MS/MS data were searched against a UniProtKB FASTA database containing 16,972 reviewed Mus musculus entries (various
dates). Propionamidation (+71.037114 @ C) was set as a fixed
modification, Deamination (+0.984016 @ N) and Acetylation
(+42.010565 @ K) were set as common1 modifications, Oxidation (+15.994915 @ M) was set as common2 modification,
and Acetylation (+42.010565 @ Protein N-term) and Methylation (+14.01565 @ K, R) were set as rare1 modifications.
Byonic was set to allow a maximum of two common modifications and one rare modification and to allow a maximum of two
missed cleavages. MS/MS spectra were matched with a tolerance of 12 ppm on precursor mass and 0.4 Da on
fragment mass.
3. Common contaminants were filtered automatically by Byonic
and include TRYP_PIG, ALBU_BOVIN, ALBU_HUMAN,
CASB_BOVIN, CASK_BOVIN, CAS1_BOVIN, CO3_HUMAN, HBA_HUMAN, HBB_HUMAN, K1M1_SHEEP,
K2C1_HUMAN,
K22E_HUMAN,
K1C10_HUMAN,
K1C15_SHEEP,
K1C9_HUMAN,
KRHB1_HUMAN,
KRHB3_HUMAN, KRHB5_HUMAN, KRHB6_HUMAN,
and TRFE_HUMAN.
4. Using Byonic, the proteome was searched with a reverse-decoy
strategy and all data were filtered and presented at a 1% false
discovery rate. Byonic calculates a Byonic score that is an
indicator of the correctness of our peptide-spectrum matches
(PSM). Byonic scores reflect the absolute quality of the PSM.
Byonic scores range from 0 to 1000, with 300 being a good
score, 400 a very good score, and scores over 500 reflecting
near-perfect matches. For our data, all filtered protein identification hits have an FDR rate lower or equal to 2.5%, a Byonic
score greater than 250, and a log probability greater than 3.
5. To further discriminate nonsignificant to statistically significant
proteins, we also recommend running a statistical power
analysis [32].
6. Once the statistically significant proteins are determined, a
spectral counting method known as normalized spectral abundance factor (NSAF) [33] can be used to determine the relative
abundance of proteins in the samples [27, 33, 34], and to verify
the reproducibility, between replicates, of the quantitative data
through a correlation analysis [34].
7. Once the statistically significant proteins are determined, along
with their relative abundance, a number of analyses can be
performed to look at functional and/or localization differences
40
Ana Gordon and Karine Gousset