Towards image-based cancer cell lines authentication using deep neural networks

Although short tandem repeat (STR) analysis is available as a reliable method for the determination of the genetic origin of cell lines, the occurrence of misauthenticated cell lines remains an important issue. Reasons include the cost, effort and time associated with STR analysis. Moreover, there are currently no methods for the discrimination between isogenic cell lines (cell lines of the same genetic origin, e.g. different cell lines derived from the same organism, clonal sublines, sublines adapted to grow under certain conditions). Hence, additional complementary, ideally low-cost and low-effort methods are required that enable (1) the monitoring of cell line identity as part of the daily laboratory routine and 2) the authentication of isogenic cell lines. In this research, we automate the process of cell line identification by image-based analysis using deep convolutional neural networks. Two different convolutional neural networks models (MobileNet and InceptionResNet V2) were trained to automatically identify four parental cancer cell line (COLO 704, EFO-21, EFO-27 and UKF-NB-3) and their sublines adapted to the anti-cancer drugs cisplatin (COLO-704rCDDP1000, EFO-21rCDDP2000, EFO-27rCDDP2000) or oxaliplatin (UKF-NB-3rOXALI2000), hence resulting in an eight-class problem. Our best performing model, InceptionResNet V2, achieved an average of 0.91 F1-score on tenfold cross validation with an average area under the curve (AUC) of 0.95, for the 8-class problem. Our best model also achieved an average F1-score of 0.94 and 0.96 on the authentication through a classification process of the four parental cell lines and the respective drug-adapted cells, respectively, on a four-class problem separately. These findings provide the basis for further development of the application of deep learning for the automation of cell line authentication into a readily available easy-to-use methodology that enables routine monitoring of the identity of cell lines including isogenic cell lines. It should be noted that, this is just a proof of principal that, images can also be used as a method for authentication of cancer cell lines and not a replacement for the STR method.

Identifikasi Isolat Antikanker Spons Hyrtios Erecta

ABSTRACTIsolation, anticancer activity test, and identification of the toxic isolate from ethanol extract of the sponge Hyrtios erecta taken from Pari Island beach (Jakarta) has conducted. Extraction of the sponges was carried out by 70% ethanol at room temperature. Partition and purification of the compounds were done by column chromatography with the stationary phase of silica gel and the mobile phase of n-hexane-chloroform (2:8). Toxicity screening test was done based on BhrineShrimp Lethality Test (BSLT). In vitro anticancer activity test of the isolate was carried out using HeLa cell line. Identification of the compounds was performed by Gas chromatography-mass spectroscopy (GC-MS). Based on theresults, it was found that the toxic isolate of H. erecta sponges has anticancer activity with IC50 of 30,497 ppm. Four compounds was detected from the anticancer isolate i.e: 4-nonylphenol; dibutyl phthalate; hexanedioic acid bis(2-ethylhexyl) ester; and cholesterol. ABSTRAKTelah dilakukan isolasi, uji aktivitas antikanker, dan identifikasi isolat toksik yang berasal dari ekstrak etanol spons Hyrtios erecta yang diambil dari perairan Pulau Pari (Jakarta). Ekstraksi dilakukan dengan cara maserasi menggunakan etanol 70% pada temperatur kamar. Pemisahan dan pemurnian komponen menggunakan kromatografi kolom dengan fase diam silikagel dan fase gerak n-heksana-kloroform (2:8). Skrining toksisitas dilakukan dengan metode Bhrine Shrimp Lethality Test (BSLT). Uji antikanker secara in vitro isolat toksik tersebut menggunakan sel HeLa. Senyawanya diidentifikasi menggunakan Gas chromatography-mass spectroscopy (GC-MS). Berdasarkan hasil penelitian ini diperoleh bahwa isolat toksik spons H. erecta bersifat antikanker dengan IC50 sebesar 30,497 ppm. Pada isolat antikankertersebut terdeteksi empat senyawa, yaitu 4-nonylphenol; dibutil phtalat; ester heksadioat bis(2-etilheksil); dan kolesterol.

Computational investigation of biological and technical variability in high throughput phenotyping and cell line identification

High-throughput cell profiling experiments are characterizing cell phenotype under a broad variety of microenvironmental and therapeutic conditions. However, biological and technical variability are contributing to wide ranges of reported parameter values, even for standard cell lines grown in identical conditions. In this paper, we develop a mathematical model of cell proliferation assays that account for biological and technical variability and limitations of the experimental platforms, including (1) cell confluency effects, (2) biological variability and technical errors in pipetting, (3) biological variability in proliferation characteristics, (4) technical variability and uncertainty in measurement timing, (5) cell counting errors, and (6) the impact of limited temporal sampling. We use this model to create synthetic datasets with growth rates and measurement times typical of cancer cell cultures, and investigate the impact of the initial cell seeding density and the common practice of fitting exponential growth curves to three cell count measurements. We find that the combined sources of variability mask the sub-exponential growth characteristics of the synthetic datasets, and that researchers profiling the same cell lines under different seeding characteristics can find significant (p < 0.05) differences in the measured growth rates. Even seeding the cells at 1% of the confluent limit can cause significant (p < 0.05) differences in the measured growth rate from the ground truth. We explored the effect of reducing errors in each part of the virtual experimental system, and found the best improvements from reducing timing errors, reducing cell counting errors, or reducing the interval between measurements (to reduce the inaccuracy of the exponential growth assumption when fitting curves). Reducing biological variability and pipetting errors had the least impact, because any improvements are still masked by cell counting errors. We close with a discussion of recommended practices for high-throughput cell phenotyping and cell line identification systems.