Attention Conference Presenters - please review the Speaker Information Page available here.
If you need assistance please contact submissions@iscb.org and provide your poster title or submission ID.
Short Abstract: Proteogenomics is a newly developing field of research, situated at the interface between genomics and proteomics. In many proteogenomic applications, mapping peptide sequences onto genome sequences can be very useful. There are many such software tools that map peptides onto genome either by taking the genomic position of peptide start site as an input or by assuming that the peptide sequence exactly matches coding sequence of the given gene model. In case of potential novel peptides resulting from genomic variations such as alternative splicing, these existing tools cannot be applied because the genome position of the novel peptide’s start site is what we really want to find out. Mapping potentially novel peptides to genome sequences, while allowing certain genomic variations, requires introducing novel gene models when aligning peptide sequences to gene structures. We have developed a new tool called ACTG (Amino aCids To Genome), which maps peptide to genome, assuming all possible single exon skipping, junction variation allowing three edit distances from the original splice sites, and frame shift. In addition, it can also consider SNVs (single nucleotide variations) during mapping phase when a user provides the VCF (variant call format) file as an input. After the mapping phase, result is reported by a flat format file containing novel genomic event information such as genomic level variations, and UTR region annotation, and also by GFF (general feature format) file containing genome positions of each mapped peptide.
Short Abstract: Integrating omics datasets become essential in scientific discovery in the post-genomics era. However, high missing rate is commonly observed in datasets generated by omics technologies, including proteomics. Imputation is often performed before many analyses take place. Motivated by studying associations between proteomics and metabolomics expression levels, we simulated 9 pairs of proteo-metabolomics full datasets with 3 different sample sizes and 3 different strengths of correlation. For each dataset, 9 missing patterns (missing at random, at low end, or mixture) with different proportions were generated. We evaluated the performance of different imputation methods in the context of proteo-metabolomics integration. In addition to “no imputation”, for each of the simulated datasets, we used the following imputation methods: minimum value, mean value, K-nearest neighbors (KNN) using peptides, KNN using samples, and probabilistic principle components analysis (PCA). Also, left censored accelerated failure time model (LAFT) and Spearman correlation were used to evaluate pairwise associations. We used false discovery rate (FDR) and true positive rate (TPR), area under the curve and accuracy to access the performance of each imputation approach. LAFT tends to have higher power but higher FDR than Spearman correlation. LAFT’s FDR is also higher than the intended controlled FDR. Imputation using overall minimum value generally outperforms other imputation methods, especially when large portion of the data is missing at low end. When the correlation is high, no imputation has reasonable performance. Consideration for sample size is very important for omics integration, especially when missing proportion is high.
Short Abstract: Drug-target identification and mechanism-of-action (MOA) studies are critical for improving drug discovery efficiency. Drug efficacy is dependent on the drug engaging the intended target while adverse effects are often caused by off-target binding and, until recently, drug-target binding events could not be measured directly in cells or tissue. The cellular thermal shift assay (CETSA), coupled with a proteomic readout, is able to detect drug-protein interactions by measuring ligand-induced thermal stability of target proteins in a cellular context. In this study, we used CETSA proteomic profiling to measure protein thermal shifts in THP-1 cells with and without treatment of the indirect NLRP3 inflammasome inhibitor CP-456,773 (aka CRID3). The data generated from these proteomics experiments were analyzed using a published analysis method [Savitski et al 2014] that includes curve fitting, normalization, estimation of slope, and calculated differences in melting points between proteins detected in treated and untreated samples. This method results in a set of proteins with statistically significant shifts in thermal stability upon compound treatment that serve as target/MOA hypotheses. Additionally, we developed and implemented an analysis method that also produces a list of proteins with statistically significant shifts in thermal stability but requires fewer analysis steps and data manipulations. Here, we present a comparison of these two methods, as well as, several MOA hypotheses.
Short Abstract: The PI3K-AKT-mTOR and the RAS-RAF-MEK-ERK pathways are commonly activated in cancer. Only a few studies have attempted to explore the spectrum of phosphorylation signaling downstream of these kinase cascades. Such insight, however, is imperative to understand the mechanisms responsible for oncogenic phenotypes.
By applying mass spectrometry-based phosphoproteomics, we mapped 5128 phosphorylation sites on 1813 proteins, and quantified their responses to activation or inhibition of PI3K, mTOR, MEK, or ERK, using isogenic knock-in derivatives and a series of targeted inhibitors. We uncovered phosphorylation changes in a wide variety of proteins involved in cell growth and proliferation, most of which have not been previously described. Multiple phosphorylation patterns revealed previously undetected feedback, convergence and crosstalk between cancer pathways, accentuating the rationale for dual pathway inhibition.
We provide a dataset rich in potential therapeutic targets downstream of the two most important cascades in cancer.
| Search Posters: |