Motivation: Identification of cell populations in flow cytometry is a critical part of the analysis and lays the groundwork for many applications and research discovery. The current paradigm of manual analysis is time consuming and subjective. A common goal of users is to replace manual analysis with automated methods that replicate their results. Supervised tools provide the best performance in such a use case, however they require fine parameterization to obtain the best results. Hence, there is a strong need for methods that are fast to setup, accurate and interpretable. Results: flowLearn is a semi-supervised approach for the quality-checked identification of cell populations. Using a very small number of manually gated samples, through density alignments it is able to predict gates on other samples with high accuracy and speed. On two state-of-the-art datasets, our tool achieves median(F1)-measures exceeding 0.99 for 31%, and 0.90 for 80% of all analyzed populations. Furthermore, users can directly interpret and adjust automated gates on new sample files to iteratively improve the initial training. Availability and implementation: FlowLearn is available as an R package on https://github.com/mlux86/flowLearn. Evaluation data is publicly available online. Details can be found in the Supplementary Material. Supplementary information: Supplementary data are available at Bioinformatics online.

BACKGROUND: Hybridization differences caused by target sequence differences can be a confounding factor in analyzing gene expression on microarrays, lead to false positives and reduce power to detect real expression differences. We prepared an R Bioconductor compatible package to detect, characterize and remove such probes in Affymetrix 3'IVT and exon-based arrays on the basis of correlation of signal intensities from probes within probe sets. RESULTS: Using completely mouse genomes we determined type 1 (false negatives) and type 2 (false positives) errors with high accuracy and we show that our method routinely outperforms previous methods. When detecting 76.2% of known SNP/indels in mouse expression data, we obtain at most 5.5% false positives. At the same level of false positives, best previous method detected 72.6%. We also show that probes with differing binding affinity both hinder differential expression detection and introduce artifacts in cancer-healthy tissue comparison. CONCLUSIONS: Detection and removal of such probes should be a routine step in Affymetrix data preprocessing. We prepared a user friendly R package, compatible with Bioconductor, that allows the filtering and improving of data from Affymetrix microarrays experiments.

MOTIVATION: When comparing gene expression levels between species or strains using microarrays, sequence differences between the groups can cause false identification of expression differences. Our simulated dataset shows that a sequence divergence of only 1% between species can lead to falsely reported expression differences for >50% of the transcripts-similar levels of effect have been reported previously in comparisons of human and chimpanzee expression. We propose a method for identifying probes that cause such false readings, using only the microarray data, so that problematic probes can be excluded from analysis. We then test the power of the method to detect sequence differences and to correct for falsely reported expression differences. Our method can detect 70% of the probes with sequence differences using human and chimpanzee data, while removing only 18% of probes with no sequence differences. Although only 70% of the probes with sequence differences are detected, the effect of removing probes on falsely reported expression differences is more dramatic: the method can remove 98% of the falsely reported expression differences from a simulated dataset. We argue that the method should be used even when sequence data are available. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.