Session: Session 4
Challenges of peptidomics-based identification of dipeptides
Bioactive peptides (2 to 20 amino acids (AAs)) exhibit a wide range of biological activities. BIOPEP is one of the most important databases of bioactive peptides with nearly 5,000 peptides referenced. A third of these latter are short-chain peptides (SCPs - composed of 2 to 6 AAs). Although analytical methods are efficient for larger peptides, SCPs still pose characterization challenges due to their “metabolomic” nature.
Firstly, 209 molecular descriptors were calculated from the SMILES of all 400 L-dipeptides and used to produce principal component analysis and a heat map to select 25 dipeptides, representative of molecular descriptor-based dipeptide heterogeneity. Each AA is present at least once in the model dipeptide set. Secondly, dipeptide analyses were carried out using a C18 HPLC-MS/MS coupled with an ESI-QTOF-MS. MS-data were then processed using Data Analysis, Metaboscape and Peaks Studio.
In this poster, we illustrate the bioinformatic strategy used to select dipeptides. Concomitantly, despite close retention times (19 dipeptides are eluted in the first minutes), manual retreatment of HPLC-MS/MS run data via Data Analysis retrieves all the 25 model peptides. The confrontation of analytical data against two Metaboscape databases, home-made from the 25 model dipeptides and from the 400 dipeptides, correctly identifies the eluted dipeptides and was unable to differentiate between peptides of the same molecular formulae, respectively. De Novo dipeptide identification using Peaks Studios resulted in the correct annotation of almost 28% of dipeptides.
Although the various dipeptides in a standard mixture can be identified manually by peptidomic condition described above, the metabolics- and peptidomics-dedicated software used does not enables accurate and automatic annotation of dipeptides. The use of other HPLC conditions (e.g., columns, temperature, …) or the use of fragments specific to each AA lead to provide more information for dipeptide identification.
Firstly, 209 molecular descriptors were calculated from the SMILES of all 400 L-dipeptides and used to produce principal component analysis and a heat map to select 25 dipeptides, representative of molecular descriptor-based dipeptide heterogeneity. Each AA is present at least once in the model dipeptide set. Secondly, dipeptide analyses were carried out using a C18 HPLC-MS/MS coupled with an ESI-QTOF-MS. MS-data were then processed using Data Analysis, Metaboscape and Peaks Studio.
In this poster, we illustrate the bioinformatic strategy used to select dipeptides. Concomitantly, despite close retention times (19 dipeptides are eluted in the first minutes), manual retreatment of HPLC-MS/MS run data via Data Analysis retrieves all the 25 model peptides. The confrontation of analytical data against two Metaboscape databases, home-made from the 25 model dipeptides and from the 400 dipeptides, correctly identifies the eluted dipeptides and was unable to differentiate between peptides of the same molecular formulae, respectively. De Novo dipeptide identification using Peaks Studios resulted in the correct annotation of almost 28% of dipeptides.
Although the various dipeptides in a standard mixture can be identified manually by peptidomic condition described above, the metabolics- and peptidomics-dedicated software used does not enables accurate and automatic annotation of dipeptides. The use of other HPLC conditions (e.g., columns, temperature, …) or the use of fragments specific to each AA lead to provide more information for dipeptide identification.