Welcome! A "Multiplexer model" is an augmented neural network that is trained from a "Base" model to provide fast, simultaneous predictions for a large set of input variations, such as all possible single nucleotide variations (SNVs) for a single sequence. Here, we provide the pre-trained BelugaMultiplexer, which predicts the effects of all possible SNVs of a single 2,000 base-pair sequence for 2,002 chromatin profiles.
In the home page, you can select the Multiplexer model and provide the corresponding input information, then submit the job to our job queue.
Chromosome - for example, chr11
Position - for example, 5255790
Target - autocomplete field and displayed in the format of Target | Cell type | Treatment, for example, type in GATA2 and select GATA2 | K562. If the treatment is missing, it will not be displayed.
The output is a plot of 4x2000 heatmap based on the sequence predictions and selected target. On top of each position of the heatmap, the heatmap also displays the reference nucleotide. You may also download the numerical predictions in pytorch serialization format (pickle-based).
Thank you for using Multiplexer. If you have any question or feedback, you can let us know at https://github.com/jzhoulab/Multiplexer/discussions