Analysing Raman spectra of crystalline cellulose degradation by fungi using artificial neural networks

Abstract

This thesis investigates the use of artificial neural networks for classifying Raman spectra of partially degraded cellulose samples by fungal species. A multilayer perceptron configuration of 4 hidden layers and 128 hidden nodes was able to classify a set of 60 samples with an overall prediction accuracy of 0.55.

Results show that data resolution is an important factor when optimizing classifier performance, and that a resolution of 1.0 cm^(-1) gave the highest performance.
We found that choosing suitable parameters for the asymmetric least squares smoothing (ALSS) correction is of relevance when attempting to optimize classifier performance, and that an ALSS smoothness value of lambda = 10^5 gave the highest performance.

Results also indicate that some fungal species and control treatments have stronger signatures in certain spectral regions. Gloeophyllum sp., Coprinellus angulatus and NaOH treatments had the most accurate probability distribution and may therefore be considered to cause the most unique cellulose modification.

This thesis shows promising results for artificial neural networks to be utilized for classifying Raman spectra of partially degraded cellulose samples.

Popular Abstract

We are surrounded by fungi. They are in the soil, on our skin and even in the air we breathe. There are millions of different fungal species on Earth, but what do they do? In nature, some fungi play an important role as decomposers, replenishing their environment with nutrients by breaking down organic matter. These "saprotrophic fungi" are capable of obtaining nutrients and energy from organic matter, such as leaves, seeds, stems, logs, roots, etc.\ and each of these species may even have their own unique set of mechanisms for doing this. Studies have shown that brown-rot and white-rot fungi do indeed differ in their methods of breaking down cellulose, but are these methods unique enough to tell them apart from other fungi? Is it possible that every fungal species has its own signature? We don’t know.

Raman spectroscopy, a technique commonly used for analysing and identifying materials, can provide spectral representations of cellulose samples which have been partially broken down by fungi. Such representations, which are associated with the molecular properties of the degraded samples, could potentially be viewed as “fungal fingerprints”. However, Raman spectroscopy can produce large amounts of data which may be too complex, or too time consuming, for humans to analyse meticulously. In such situations, artificial neural networks can be utilized.

Artificial neural networks can be described as computational models, vaguely inspired by the human brain, capable of acquiring and maintaining information. These networks can easily be provided the spectral data of the cellulose samples which they will learn to associate with their corresponding species, all by themselves. If training is successful, a network will be able to correctly identify a sample it has never seen before. Artificial neural networks may therefore provide a tool powerful enough to further help us understand the similarities and differences between fungal species, as well as their mechanisms.

• author: Hutteri, Sebastian
• supervisor: Carl Troein
• organization: Computational Biology and Biological Physics - Undergoing reorganization; Department of Astronomy and Theoretical Physics - Undergoing reorganization
• course: FYTK02 20182
• year: 2019
• type: M2 - Bachelor Degree
• subject: Physics and Astronomy
• keywords: Raman spectroscopy, artificial neural networks, saprotrophic fungi, cellulose
• language: English
• id: 8972265
• date added to LUP: 2019-03-27 14:06:15
• date last changed: 2019-06-19 11:28:14