| 1. |
- Basile, Walter, 1980-
(författare)
-
Orphan Genes Bioinformatics : Identification and properties of de novo created genes
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Even today, many genes are without any known homolog. These "orphans" are found in all species, from Viruses to Prokaryotes and Eukaryotes. For a portion of these genes, we might simply not have enough data to find homologs yet. Some of them are imported from taxonomically distant organisms via lateral transfer; others have homologs, but mutated beyond the point of recognition.However, a sizeable fraction of orphan genes is unambiguously created via "de novo" mechanisms. The study of such novel genes can contribute to our understanding of the emergence of functional novelty and the adaptation of species to new ecological niches.In this work, we first survey the field of orphan studies, and illustrate some of the common issues. Next, we analyze some of the intrinsic properties of orphans proteins, including secondary structure elements and Intrinsic Structural Disorder; specifically, we observe that in young proteins the relationship between these properties and the G+C content of their coding sequence is stronger than in older proteins.We then tackle some of the methodological problems often found in orphan studies. We find that using evolutionarily close species, and sensitive, state-of-the art homology recognition methods is instrumental to the identification of a set of orphans enriched in de novo created ones.Finally, we compare how intrinsic disorder is distributed in bacteria versus eukaryota. Eukaryotic proteins are longer and more disordered; the difference is to be attributed primarily to eukaryotic-specific domains and linker regions. In these sections of the proteins, a higher frequency of the disorder-promoting amino acid Serine can be observed in Eukaryotes.
|
|
| 2. |
- Michel, Mirco, 1986-
(författare)
-
From Sequence to Structure : Using predicted residue contacts to facilitate template-free protein structure prediction
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Despite the fundamental role of experimental protein structure determination, computational methods are of essential importance to bridge the ever growing gap between available protein sequence and structure data. Common structure prediction methods rely on experimental data, which is not available for about half of the known protein families.Recent advancements in amino acid contact prediction have revolutionized the field of protein structure prediction. Contacts can be used to guide template-free structure predictions that do not rely on experimentally solved structures of homologous proteins. Such methods are now able to produce accurate models for a wide range of protein families.We developed PconsC2, an approach that improved existing contact prediction methods by recognizing intra-molecular contact patterns and noise reduction. An inherent problem of contact prediction based on maximum entropy models is that large alignments with over 1000 effective sequences are needed to infer contacts accurately. These are however not available for more than 80% of all protein families that do not have a representative structure in PDB. With PconsC3, we could extend the applicability of contact prediction to families as small as 100 effective sequences by combining global inference methods with machine learning based on local pairwise measures.By introducing PconsFold, a pipeline for contact-based structure prediction, we could show that improvements in contact prediction accuracy translate to more accurate models. Finally, we applied a similar technique to Pfam, a comprehensive database of known protein families. In addition to using a faster folding protocol we employed model quality assessment methods, crucial for estimating the confidence in the accuracy of predicted models. We propose models tobe accurate for 558 families that do not have a representative known structure. Out of those, over 75% have not been reported before.
|
|
| 3. |
- Tsirigos, Konstantinos, 1982-
(författare)
-
Bioinformatics Methods for Topology Prediction of Membrane Proteins
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Membrane proteins are key elements of the cell since they are associated with a variety of very important biological functions crucial to its survival. They are implicated in cellular recognition and adhesion, act as molecular receptors, transport substrates through membranes and exhibit specific enzymatic activity.This thesis is focused on integral membrane proteins, most of which contain transmembrane segments that form an alpha helix and are composed of mainly hydrophobic residues, spanning the lipid bilayer. A more specialized and less well-studied case, is the case of integral membrane proteins found in the outer membrane of Gram-negative bacteria and (presumably) in the outer envelope of mitochondria and chloroplasts, proteins whose transmembrane segments are formed by amphipathic beta strands that create a closed barrel (beta-barrels). The importance of transmembrane proteins, as well as the inherent difficulties in crystallizing and obtaining three-dimensional structures of these, dictates the need for developing computational algorithms and tools that will allow for a reliable and fast prediction of their structural and functional features. In order to elucidate their function, we must acquire knowledge about their structure and topology with relation to the membrane. Therefore, a large number of computational methods have been developed in order to predict the transmembrane segments and the overall topology of transmembrane proteins. In this thesis, I initially describe a large-scale benchmark of many topology prediction tools in order to devise a strategy that will allow for better detection of alpha-helical membrane proteins in a proteome. Then, I give a description of construction of improved machine-learning algorithms and computer software for accurate topology prediction of transmembrane proteins and discrimination of such proteins from non-transmembrane proteins. Finally, I introduce a fast way to obtain a position-specific scoring matrix, which is essential for modern topology prediction methods.
|
|
| 4. |
- Uziela, Karolis, 1987-
(författare)
-
Protein Model Quality Assessment : A Machine Learning Approach
- 2017
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Many protein structure prediction programs exist and they can efficiently generate a number of protein models of a varying quality. One of the problems is that it is difficult to know which model is the best one for a given target sequence. Selecting the best model is one of the major tasks of Model Quality Assessment Programs (MQAPs). These programs are able to predict model accuracy before the native structure is determined. The accuracy estimation can be divided into two parts: global (the whole model accuracy) and local (the accuracy of each residue). ProQ2 is one of the most successful MQAPs for prediction of both local and global model accuracy and is based on a Machine Learning approach.In this thesis, I present my own contribution to Model Quality Assessment (MQA) and the newest developments of ProQ program series. Firstly, I describe a new ProQ2 implementation in the protein modelling software package Rosetta. This new implementation allows use of ProQ2 as a scoring function for conformational sampling inside Rosetta, which was not possible before. Moreover, I present two new methods, ProQ3 and ProQ3D that both outperform their predecessor. ProQ3 introduces new training features that are calculated from Rosetta energy functions and ProQ3D introduces a new machine learning approach based on deep learning. ProQ3 program participated in the 12th Community Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP12) and was one of the best methods in the MQA category. Finally, an important issue in model quality assessment is how to select a target function that the predictor is trying to learn. In the fourth manuscript, I show that MQA results can be improved by selecting a contact-based target function instead of more conventional superposition based functions.
|
|
| 5. |
- Peters, Christoph, 1983-
(författare)
-
Topology Prediction of α-Helical Transmembrane Proteins
- 2016
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Membrane proteins fulfil a number of tasks in cells, including signalling, cell-cell interaction, and the transportation of molecules. The prominence of these tasks makes membrane proteins an important target for clinical drugs. Because of the decreasing price of sequencing, the number of sequences known is increasing at such a rate that manual annotations cannot compete. Here, topology prediction is a way to provide additional information. It predicts the location and number of transmembrane helices in the protein and the orientation inside the membrane. An important factor to detect transmembrane helices is their hydrophobicity, which can be calculated using dedicated scales. In the first paper, we studied the difference between several hydrophobicity scales and evaluated their performance. We showed that while they appear to be similar, their performance for topology prediction differs significantly. The better performing scales appear to measure the probability of amino acids to be within a transmembrane helix, instead of just being located in a hydrophobic environment.Around 20% of the transmembrane helices are too hydrophilic to explain their insertion with hydrophobicity alone. These are referred to as marginally hydrophobic helices. In the second paper, we studied three of these helices experimentally and performed an analysis on membrane proteins. The experiments show that for all three helices positive charges on the N-terminal side of the subsequent helix are important to insert, but only two need the subsequent helix. Additionally, the analysis shows that not only the N-terminal helices are more hydrophobic, but also the C-terminal transmembrane helices.In Paper III, the finding from the second paper was used to improve the topology prediction. By extending our hidden Markov model with N- and C-terminal helix states, we were able to set stricter cut-offs. This improved the general topology prediction and in particular miss-prediction in large N- and C-terminal domains, as well the separation between transmembrane and non-transmembrane proteins.Lastly, we contribute several new features to our consensus topology predictor, TOPCONS. We added states for the detection of signal peptides to its hidden Markov model and thus reduce the over-prediction of transmembrane helices. With a new method for the generation of profile files, it is possible to increase the size of the database used to find homologous proteins and decrease the running time by 75%.
|
|
| 6. |
- Salvatore, Marco, 1988-
(författare)
-
Predicting the route: from protein sequence to sorting in eukaryotic cell
- 2019
-
Doktorsavhandling (övrigt vetenskapligt/konstnärligt)abstract
- Proteins need to be localised in the correct compartment of a eukaryotic cell to function correctly. Therefore, a protein needs to be transported to the right location. Specific signals present in the protein sequence direct proteins to different subcellular localisations. The correct transport is essential for the life of the cell, while, possible errors during the transport can cause irreversible damage and interfere with the activities of surrounding proteins. For more than 30 years, the development of methods to identify the localisation of proteins using both experimental and computational approaches has been an important research area. The objective of this thesis is to develop better computational methods for the classification of the subcellular localisation of eukaryotic proteins. I first describe the development of a consensus method, SubCons, which improves the subcellular prediction of human proteins. Next, I present the SubCons web-server as well as an additional benchmark using protein annotation from novel mass-spectrometry studies in two eukaryotic organisms Mus musculus and Drosophila melanogaster. Then, I present the new version of TargetP and how deep learning can improve the identification of N-terminal sorting signals by focusing on relevant biological signatures. Finally, I describe the development of a novel method for sub-nuclear localisation prediction. Here, I show that the performance of a deep convolutional neural network is improved when using an augmented dataset of homologous proteins.
|
|
