What does it mean for an algorithm to be biased? In U.S. law, unintentional bias is encoded via disparate impact, which occurs when a selection process has widely different outcomes for different groups, even as it appears to be neutral. This legal determination hinges on a definition of a protected class (ethnicity, gender) and an explicit description of the process.When computers are involved, determining disparate impact (and hence bias) is harder. It might not be possible to disclose the process. In addition, even if the process is open, it might be hard to elucidate in a legal setting how the algorithm makes its decisions. Instead of requiring access to the process, we propose making inferences based on the data it uses.We present four contributions. First, we link disparate impact to a measure of classification accuracy that while known, has received relatively little attention. Second, we propose a test for disparate impact based on how well the protected class can be predicted from the other attributes. Third, we describe methods by which data might be made unbiased. Finally, we present empirical evidence supporting the effectiveness of our test for disparate impact and our approach for both masking bias and preserving relevant information in the data. Interestingly, our approach resembles some actual selection practices that have recently received legal scrutiny.

During protein synthesis, the ribosome controls the movement of transfer RNA (tRNA) and messenger RNA (mRNA) by means of large-scale structural rearrangements. We describe structures of the intact bacterial ribosome from Escherichia coli that reveal how the ribosome binds tRNA in two functionally distinct states, determined to a resolution of ~3.2 Å by x-ray crystallography. One state positions tRNA in the peptidyl-tRNA binding site. The second, a fully rotated state, is stabilized by ribosome recycling factor (RRF) and binds tRNA in a highly bent conformation in a hybrid peptidyl/exit (P/E) site. The structures help to explain how the ratchet-like motion of the two ribosomal subunits contributes to the mechanisms of translocation, termination, and ribosome recycling.

Aminoacyl-tRNA (aa-tRNA), in a ternary complex with Elongation Factor-Tu (EF-Tu) and GTP, enters the aminoacyl (A) site of the ribosome via a multi-step, mRNA codon-dependent mechanism. This process gives rise to the preferential selection of cognate aa-tRNAs for each mRNA codon and consequently the fidelity of gene expression. The ribosome actively facilitates this process by recognizing structural features of the correct substrate, initiated in its decoding site, to accelerate the rates of EF-Tu-catalyzed GTP hydrolysis and ribosome-catalyzed peptide bond formation. Here, the order and timing of conformational events underpinning the aa-tRNA selection process were investigated from multiple structural perspectives using single-molecule fluorescence resonance energy transfer (smFRET). The time resolution of these measurements was extended to 2.5 and 10ms, a 10–50-fold improvement over previous studies. The data obtained reveal that aa-tRNA undergoes fast conformational sampling within the A site, both before and after GTP hydrolysis. This suggests that the alignment of aa-tRNA with respect to structural elements required for irreversible GTP hydrolysis and peptide bond formation plays a key role in the fidelity mechanism. These observations provide direct evidence that the selection process is governed by motions of aa-tRNA within the A site, adding new insights into the physical framework that helps explain how the rates of GTP hydrolysis and peptide bond formation are controlled by the mRNA codon and other fidelity determinants within the system.