Teng, H. | bioRxiv Science

bioRxiv2018

Decreased green autofluorescence intensity of lung parenchyma is a potential non-invasive diagnostic biomarker for lung cancer

Early and non-invasive diagnosis is critical for enhancing the survival rates of lung cancer. In current study we determined the green autoflorescence (AF) of the pulmonary parenchyma of lung cancer patients, using 488 nm and 500 - 550 nm as excitation wavelength and emission wavelength, respectively. Our study has suggested that decreased green AF intensity of pulmonary parenchyma may be a potential diagnostic biomarker for lung cancer: First, the green AF intensity of the cancerous tissues is less than 40% of those of distant non-neoplastic tissues and peri-neoplastic tissues; second, the green AF intensity of both the distant non-neoplastic tissues and cancerous tissues of squamous carcinoma is significantly lower than that of adenocarcinoma; third, the AF intensity of the peri-neoplastic tissues of the lung cancer patients is negatively correlated with the stages of lung cancer; and fourth, our study on the AF spectrum of pulmonary parenchyma has suggested that the AF may result from the AF of the keratins or FAD of the pulmonary parenchyma. Collectively, our study has suggested that decreased green AF intensity of pulmonary parenchyma may become a novel biomarker for non-invasive diagnosis of lung cancer. The green AF intensity may also be used to non-invasively differentiate squamous carcinoma and adenocarcinoma, as well as the stages of lung cancer. Moreover, our diagnostic approach for lung cancer may be used for image-guided surgery of lung cancer for lesions that have not been identified by CT.

biophysics↗

bioRxiv2017

Chiron: Translating nanopore raw signal directly into nucleotide sequence using deep learning

Sequencing by translocating DNA fragments through an array of nanopores is a rapidly maturing technology which offers faster and cheaper sequencing than other approaches. However, accurately deciphering the DNA sequence from the noisy and complex electrical signal is challenging. Here, we report Chiron, the first deep learning model to achieve end-to-end basecalling: directly translating the raw signal to DNA sequence without the error-prone segmentation step. Trained with only a small set of 4000 reads, we show that our model provides state-of-the-art basecalling accuracy even on previously unseen species. Chiron achieves basecalling speeds of over 2000 bases per second using desktop computer graphics processing units.

bioinformatics↗